THE PREDICTIVE VALIDITY OF A PSYCHOLOGICAL TEST BATTERY FOR

THE SELECTION OF CADET PILOTS IN A COMMERCIAL AIRLINE

by

VUSUMUZI QUIRION MNGUNI

submitted in accordance with the requirements for

the degree of

MASTER OF COMMERCE

in the subject

INDUSTRIAL AND ORGANISATIONAL PSYCHOLOGY

at the

UNIVERSITY OF SOUTH AFRICA

SUPERVISOR: PROF. M. DE BEER

MARCH 2011

iii.

STATEMENT: I declare that this dissertation: “The predictive validity of a psychological test

battery for the selection of cadet pilots in a commercial airline”, is my own work

and that all the resources used or quoted have been indicated and acknowledged by

means of complete reference.

________________ ___________________

V. Q. MNGUNI DATE

ACKNOWLEDGEMENTS

I would like to express my gratitude to the following people: South African Airways management and staff for allowing and giving me access to the information for this research. The Almighty for giving me the opportunity and grace to undertake this research until completion. Frans van Staden for his encouragement, support and believing in my capability. My wife Pheello and children for their understanding and support as well as sacrifice of their time. Finally, my sincere appreciation and gratitude to my supervisor Prof. M. De Beer, for her guidance, assistance, reassurance and encouragement, which has made me achieve a milestone in completing this research.

ii

xv

SUMMARY

THE PREDICTIVE VALIDITY OF A PSYCHOLOGICAL TEST

BATTERY FOR THE SELECTION OF CADET PILOTS FOR A

COMMERCIAL AIRLINE

By Vusumuzi Quirion Mnguni Degree: M. Comm. Subject: Industrial Psychology Supervisor: Prof. M. de Beer

Commercial airlines need to employ well qualified pilots to run their core business. The current

supply from privately and military qualified pilots is proving to be inadequate. A further challenge

facing the airline is having to attempt to reflect the diversity of the country in its workforce.

The present study investigated the predictive validity of a psychological test battery for cadet

pilots. The predictors that were included in the research are: biographical data, ABET levels in

terms of English and Matric results, as well as results from psychological tests, namely: English

literacy skills assessment (ELSA), Raven’s Progressive Matrices (RMP), Blox test, subtests of

the Intermediate Battery (B/77) viz: Arithmetic 1 and 2, and Reading Comprehension, and the

Wechsler Adult Intelligence Scale (WAIS). The objective of the research was to determine the

predictive validity of the selection battery utilizing the final flying school results as the criteria.

The results of the research were inconclusive. Only some of the tests showed positive

correlations with the modules of the flying school results. The Ravens Progressive Matrices,

Blox, Matriculation English symbol, ABET levels and Reading Comprehension, were found to

have predictive power with some of the modules of the flying school results based on the

regression analysis.

It is recommended that a revised profile for a commercial airline pilot should be developed, as

well as that the critical skills and competencies should be identified to enable the airlines to

utilize appropriate and relevant assessment tools to select prospective candidates, particularly

among the previously disadvantaged communities.

xvi

Key words

i. Psychological tests; intelligence/aptitude; literacy; numeracy; predictive validity;

selection; cadet pilot.

iv

TABLE OF CONTENTS

Page ACKNOWLEDGEMENTS ii LIST OF TABLES xi LIST OF FIGURES xiv SUMMARY xv CHAPTER 1

BACKGROUND TO AND RATIONALE FOR THE RESEARCH 1 1.1 INTRODUCTION 1

1.2 PROBLEM STATEMENT 6

1.3 RESEARCH AIMS 7

1.3.1 General aim 7

1.3.2 Specific aims of the literature review 8

1.3.3 Specific aims of the empirical study 8

1.4 THE PARADIGM PERSPECTIVE OF THE RESEARCH 8

1.4.1 Applicable psychological paradigms 8

1.4.1.1 Cognitive behaviourism

1.4.1.2 Functionalism

1.4.1.3 Empiricism

1.4.2 Meta-theoretical statements 10

1.4.3 Market of intellectual resources 12

1.4.4 Psychological models and theories 13

1.4.5 Applicable concepts and constructs 13

1.4.6 Methodological assumptions 15

1.5 RESEARCH DESIGN 16

1.5.1 Research variables 16

v

1.5.1.1 Independent variables 16

1.5.1.2 Dependent variables 17

1.6 RESEARCH METHOD 17

1.6.1 Literature review 17

1.6.1.1 The use of psychological tests in the selection of cadet pilots 17

1.6.1.2 The nature of literacy and numeracy 17

1.6.1.3 Integration 18

1.6.2 The empirical study 18

1.6.2.1 Data collection 18

1.6.2.2 Data processing 18

1.6.2.3 Research hypotheses 18

1.6.2.4 Reporting the results 19

1.6.2.5 Conclusion, limitations and recommendations 19

1.7 LAYOUT OF CHAPTERS 19

1.8 CHAPTER SUMMARY 20

CHAPTER 2 THE USE OF PSYCHOLOGICAL TESTS IN SELECTION 21

WITH SPECIFIC FOCUS ON MEASUREMENTS OF COGNITIVE

ASPECTS

2.1 INTRODUCTION 21

2.2 DEFINITION OF CONCEPTS 21

2.2.1 Selection process 22

2.2.2 Definition of selection 25

2.3 PSYCHOLOGICAL TESTS 26

2.3.1 Definition of psychological tests 27

2.3.2 Functions of psychological tests 28

2.4 THE DETERMINANTS OF PILOT TRAINING SUCCESS 32

2.4.1 Intelligence and aptitude 32

2.4.2 Psychomotor coordination 32

vi

2.4.3 Personality 33

2.5 THE DEVELOPMENT AND THEORIES OF INTELLIGENCE 33

2.5.1 Spearman’s two-factor model 34

2.5.2 Thurstone’s primary mental abilities model 34

2.5.3 Cattell’s model 35

2.5.4 Gustafsson’s three-level model 37

2.5.5 Carroll’s three-stratum theory 38

2.5.6 Definition of intelligence 40

2.5.7 Measurement of intelligence 43

2.6 DEFINITION OF APTITUDE 45

2.6.1 Measurement of aptitude 50

2.7 PSYCHOLOGICAL TESTS THAT MEASURE INTELLIGENCE 53

AND APTITUDE

2.7.1 Requirements for psychological measures 58

2.7.1.1 Reliability 58

2.7.1.2 Validity 59

2.8 OVERVIEW OF THE STRUCTURE AND FORMAT 61

OF THE CADET PILOT TRAINING PROGRAMME

2.8.1 Content of the cadet training programme 62

2.8.2 Duties and responsibilities of a pilot 63

2.8.3 Psychological profile of a pilot 65

2.9 CHAPTER SUMMARY 66

CHAPTER 3 LITERACY AND NUMERACY 67

3.1 INTRODUCTION 67

3.2 LITERACY 67

3.2.1 Literacy in the South African context 68

3.2.2 Definition of literacy 69

3.2.3 Development of language skills for basic English 72

vii

3.2.3.1 Theories and models in language testing 72

3.2.3.2 Language testing 74

3.2.3.3 Language skills for basic English literacy 79

3.2.4 Components of language 81

3.2.4.1 Context of language 82

3.2.4.2 Forms of language 82

3.2.4.3 Use of language 83

3.2.5 Basic dimensions of language 84

3.2.5.1 Basic functions of language 85

3.3 NUMERACY 94 3.3.1 Definition of numeracy 94

3.3.2 Development of skills for basic numeracy 96

3.3.2.1 Theories, models and skills for basic numeracy 96

3.3.3 Basic dimensions of numeracy 100

3.3.3.1 Basic functions of numeracy 101

3.4 INTEGRATION OF LANGUAGE, LITERACY AND NUMERACY 104 3.5 CHAPTER SUMMARY 107 CHAPTER 4 THE EMPIRICAL STUDY 108

4.1 INTRODUCTION 108

4.2 DESCRIPTION OF THE POPULATION AND SAMPLE 108

4.2.1 Biographical profile of the sample in the second phase (phase 2) 112

4.3 PSYCHOLOGICAL TESTS USED AS MEASURING 117 INSTRUMENTS 4.3.1 General proposed profile of a pilot 117

4.3.2 Psychological assessment instruments used in the cadet pilot 122

selection procedure

4.3.2.1 English Literacy Skills Assessment (ELSA) 122

4.3.2.2 The Blox Test 125

4.3.2.3 Raven’s Progressive Matrices (RPM) test 127

viii

4.3.2.4 The Intermediate Battery (B/77) 130

4.3.2.5 The Wechsler Adult Intelligence Scale (WAIS) 132

4.3.2.6 Biographical factors and matriculation results 139 4.4 DATA COLLECTION AND PROCESSING 140 4.4.1 Data collection for the independent variables 141

4.4.2 Data collection for the dependent variables 141 4.5 DATA ANALYSIS 142

4.5.1 Introduction 142

4.5.2 Descriptive statistics 142

4.5.3 Inferential statistics 143

4.5.3.1 ANOVA 144

4.5.3.2 Correlation analysis 144

4.5.3.3 Regression analysis 145

4.5.4 Statistical techniques used 146

4.5.5 Level of statistical significance 147

4.5.6 Practical effect size 148

4.6 RESEARCH HYPOTHESES 149 4.6.1 Hypotheses regarding the validity of each psychological test 149 in phase 1 of the selection for predicting performance in a cadet pilot training programme 4.6.2 Hypotheses regarding the validity of each psychological test 150 in phase 2 of the selection for predicting performance in a cadet pilot training programme 4.7 CHAPTER SUMMARY 151 CHAPTER 5

THE RESEARCH RESULTS 152

5.1 INTRODUCTION 152 5.2 RESULTS OF THE PSYCHOMETRIC TESTS IN 152 PHASE 1: COMPARISON BETWEEN THE RACE AND GENDER GROUPS

5.2.1 Results of the tests per race group on the independent variables 154

5.2.1.1 Summary 158

5.2.2 Results for the tests by gender (within race) 159

ix

5.2.2.1 Blacks 160

5.2.2.2 Coloureds 162

5.2.2.3 Asians 164

5.2.2.4 Whites 166

5.2.2.5 Summary of descriptive results 167

5.2.3 Relationship between psychometric 167 tests (phase 1)

5.3 THE RESULTS OF THE PSYCHOMETRIC SUBTESTS DURING 168 PHASE 2: COMPARISON OF RACE GROUPS 5.4 SUMMARY OF THE COMPARISON OF THE SUBGROUP RESULTS 175 5.5 CORRELATION BETWEEN PHASE 1 AND PHASE 2 176

RESULTS AND PERFORMANCE IN THE CADET PILOT

TRAINING PROGRAMME

5.5.1 Correlations between the phase 1 test results and performance 177

in the cadet training programme

5.5.2 Correlations between the phase 2 test results and performance 182

in the cadet training programme

5.5.3 Stepwise regression analysis 188

5.6 INTEGRATION OF RESULTS 196

5.6.1 Descriptive statistics and group comparison 196

5.6.2 Correlation analysis 196

5.6.3 Stepwise regression analysis 197

5.7 TESTING OF THE RESEARCH HYPOTHESES 197

5.7.1 Hypothesis testing regarding the psychological tests 199

5.7.2 Discussion 204

5.8 CHAPTER SUMMARY 207 CHAPTER 6

CONCLUSIONS, LIMITATIONS AND RECOMMENDATIONS FOR 208 FUTURE RESEARCH 6.1 INTRODUCTION 208

6.1.1 General conclusions 208

x

6.1.2 Specific conclusions relating to the literature review 209

6.1.2.1 Conclusions relating to intelligence, aptitude, psychological tests 209

and selection

6.1.2.2 Conclusions relating to literacy and numeracy 210

6.1.3 Specific conclusions relating to the empirical research 211

6.1.3.1 Conclusions relating to psychological test battery 211

6.1.3.2 Conclusions relating to the importance of other variables: English 212

Matric symbol and literacy as well as ABET levels, in determining

the successful completion of the cadet training programme

6.1.4 Potential resolution of the research problem 212 6.2 LIMITATIONS OF THE PRESENT RESEARCH 213

6.2.1 Practical constraints 214

6.2.2 Restriction of range 214

6.2.3 Sample 215

6.3 RECOMMENDATIONS 216

6.3.1 Incorporation of other psychological paradigms 216

6.3.2 Cross-validation 218

6.3.3 Longitudinal study of progress through performance 218

6.4 CHAPTER SUMMARY 219 REFERENCES 221

xi

LIST OF TABLES Table 2.1: Content of each module in phase 1: the preparation course 62 Table 2.2: Content of each module in phase 2: the flying course 63 Table 2.3: General duties of pilot 64 Table 3.1: The four basic components of comprehension 91 Table 3.2: Differences between arithmetic and numeracy 95 Table 4.1: Distribution of applications by half yearly intake 109 Table 4.2: Race distribution of applicants 110 Table 4.3: Gender distribution of applicants 110 Table 4.4: Mid-year estimates for South Africa by population group 111 and sex, 2006 Table 4.5: Race by gender distribution of applicants 112 Table 4.6: Race distribution of second phase candidates 113 Table 4.7: Gender distribution of second phase candidates 114 Table 4.8: Race and gender of second phase candidates 115 Table 4.9: Proportion of each demographic group included 116

in phase 2, as a percentage of phase 1 participants Table 4.10: Proposed profile of a pilot 117 Table 4.11: The psychological instruments used 121 Table 5.1: Descriptive statistics for the psychometric tests in phase 1 154 Table 5.2: ANOVA comparison of race groups in each of the tests 156 Table 5.3: Scheffé subsets of race group on Raven’s Test 157 Table 5.4: Scheffé subsets of race group on the Blox Test 157 Table 5.5: Scheffé subsets of race group on the reading 158 comprehension test Table 5.6: Scheffé subsets of race group in arithmetic 1 158 Table 5.7: Scheffé subsets of race group on arithmetic 2 158

xii

Table 5.8: Descriptive statistics for gender among blacks on 160 the independent variables Table 5.9: ANOVA comparison of the black gender groups 161 Table 5.10: Descriptive statistics per gender group for the coloureds 162 in the independent variables Table 5.11: ANOVA comparison of the coloured gender groups 163 Table 5.12: Descriptive statistics per Asian gender group on the 164 independent variables Table 5.13: ANOVA comparison of the Asian gender groups 165 Table 5.14: Descriptive statistics per the white gender group 166 on the independent variables Table 5.15: ANOVA comparison of the white gender groups 167 Table 5.16: Correlation matrix of the tests in phase 1 168 Table 5.17: Descriptive statistics for the test results for phase 2, 169 per gender group Table 5.18: Tests of between-subjects’ effects per test between race groups 171 Table 5.19: Scheffé test results for level of literacy 172 Table 5.20: Scheffé test results for the ABET levels 173 Table 5.21: Scheffé results for the English Matric symbol 173 Table 5.22: Scheffé results for digits combined 174 Table 5.23: Scheffé results for picture completion 174 Table 5.24: Scheffé results for object assembly 175 Table 5.25: Scheffé results for block design 175 Table 5.26: Correlations between the phase 1 test results and 178 the final flying school results Table 5.27: Correlations between the phase 2 test results and 183 the final flying school results Table 5.28: Predicting performance in meteorology 190 Table 5.29: Predicting performance in navigation 192 Table 5.30: Predicting performance in radio 193

xiii

Table 5.31: Predicting performance in air law and procedures 193 Table 5.32: Predicting performance in human performance 194 Table 5.33: Predicting performance in aircraft technical 195 Table 5.34: Hypothesis testing of the results 198

xiv

LIST OF FIGURES Figure 2.1: The selection process 24 Figure 2.2: Gustafsson’s hierarchical model of intelligence 37 Figure 3.1: The three bases of the environment as seen from 87 the perspective of language Figure 3.2: Components of reading 88 Figure 3.3: Numeracy in a social context 99 Figure 3.4: Model integrating the development of the 104

numeracy concepts, skills and language Figure 5.1: Mean score per race group for the psychometric tests in phase 1 155

1

CHAPTER 1

BACKGROUND TO AND RATIONALE FOR THE RESEARCH

1.1 INTRODUCTION

According to Marais (2010), the official unemployment rate in South Africa rose to 23.5% in the

first quarter of 2010, from 21.9% in the previous quarter, ending a five-year decline. He further

reports that the actual unemployment rate is closer to 40%, and among young African men and

women it probably exceeds 60%. This information emphasises the fact that the country needs to

rebuild local industrial capacity. The government has promised to create four million “new jobs”

by 2014 (Marais, 2010). It can be inferred that there is currently an oversupply of applicants

chasing the few available jobs, which poses a challenge for organisations and employers who

need to appoint employees with a better chance of succeeding in their positions.

For years, the airline industry has been perceived as a white male-dominated industry.

Vermeulen (2009) concurs and reports that in the last decade, new South African government

and labour policies have been put in place to encourage more women to become aviators.

However, even though women have been aviators for almost as long as men, aviation is still

largely a male-dominated profession.

According to Copans (2007), Frankel (president of the Commercial Aviation Association of

South Africa [CAASA]) has noted that historically, many airline pilots were originally air force

pilots. However, this is no longer the case, and there is now a dearth of skilled pilots in the

aviation industry. Hence substantial effort is required to attract more people to the aviation

industry, including previously disadvantaged people, so that they can be trained as pilots.

According to Frankel (Copans, 2007), the increased inclusion of previously disadvantaged

people into aerospace industry is imperative for the sustainability of the industry. He further

asserts that aggressive, government-supported training and apprentice schemes are required to

attract young black talent to this segment of the industry.

According to Theron (2009), selection from a diverse group poses a real and formidable

challenge to the field of industrial psychology in South Africa. The challenge is to develop

2

selection procedures that simultaneously add value, do not discriminate unfairly and minimise

the adverse impact. Theron (2009) emphasised that the success of organisations in such an

undertaking of efficiently combining and transforming scarce factors of production into products

and services with economic utility, requires competent high-performing employees. At the same

time, however, South African organisations are under moral, economic, political and legal

pressure to diversify their workforce.

It can be argued that the promulgation of legislation, particularly the Employment Equity Act 55

of 1998 (EEA) and the Skills Development Act 97 of 1998, after the 1994 democratic elections,

introduced a number of new employment practices.

Psychometric assessment in most South African organisations was part and parcel of the

selection process. However, the promulgation of the above-mentioned Acts has prompted

organisations to review their employment processes and procedures with a particular focus on

the use of, inter alia, psychological assessments.

According to Muller and Schepers (2003) and Van de Vijver and Rothmann (2004), in South

Africa, reference to psychometric testing is made in the EEA, chapter 2 point 8: “Psychometric

testing of an employee is prohibited unless the test -

a) Has been scientifically validated as providing reliable results which are appropriate for

the intended purpose;

b) Can be applied fairly to employees irrespective of their culture, and

c) Is not biased against people from designated groups.

According to this legislation, the burden of proof has been placed on the organisation to justify

the utilisation of psychological instruments in their employment practices with the emphasis on

the previously disadvantaged people. This view is supported by Nzama, De Beer and Visser

(2008) who confirmed that the introduction of the Labour Relations Act (1995) has compelled

organisations to ensure that their employment practices are fair.

These views indicate that the selection instruments and procedures need to be re-evaluated to

ensure that they discriminate fairly. Theron (2009) refers to the Uniform Guideline on Employee

Selection Procedures published by the Equal Employment Opportunity Commission (EEOC)

which endorses this position by stipulating the following: “Where two or more selection

3

procedures are available which serve the user’s legitimate interest in efficient and trustworthy

workmanship, and which are substantially equally valid for a given purpose, the user should use

the procedure which has been demonstrated to have the lesser adverse impact” (EEOC, 1978,

p. 3897).

Owing to the oversupply of job applicants, one of the functions of human resources is to select

and appoint employees who will help the organisation succeed by selecting, appointing and

retaining a competent and motivated workforce. Theron (2009) concurred when he postulated

that the objective of personnel selection is to allow only those applicants to enter the

organisation who would perform satisfactorily in their designated positions.

Another challenge for organisations and employers is the Skills Development Act 97 of 1998),

the purpose of which is

a) To develop the skills of South African workforce;

b) To increase the levels of investment in education and training in the labour market and

improve the return on that investment;

c) To encourage employers -

i. To use the workplace as an active learning environment;

ii. To provide employees with the opportunities to acquire new skills;

iii. To provide opportunities for new entrants to the labour market to gain work

experience; and

iv. To employ persons who find it difficult to be employed

d) To encourage workers to participate in learnership and other training programmes;

e) To improve the employment prospects of person previously disadvantaged by unfair

discrimination and redress those disadvantages through training and education;

f) To ensure the quality of education and training in and for the workplace

g) To assist -

i. Work-seekers to find work;

ii. Retrenched workers to re-enter the labour market;

iii. Employers to find qualified employees; and

h) To provide and regulate employment services.

This legislation has placed another burden on organisations, namely to create opportunities to

uplift the standard of living of historically disadvantaged people. According to Meiring, Van de

4

Vijver, Rothmann and Barrick (2005), there is an urgent need for measuring instruments that

meet the EEA requirements and that can be used for all the culture and language groups in

South Africa.

Van Der Merwe (2002) stated that one of the critical elements in ensuring outstanding

organisational performance is the selection and development of excellent staff. International as

well as local research has demonstrated the role that psychometric assessment can play in

significantly improving the selection process for both new entrants and internal promotions.

According to Van Der Merwe (2002), effective psychometric assessment can also play a part in

staff development processes – a critical challenge that currently faces South Africa.

Paterson and Uys (2005) have noted that the changing world of work has an impact on

assessment practices. They suggest that in the “new” organization, the focus is on recruiting

and developing employees with the ability to work flexibly and adaptively, owing to rapid

changes inside and outside the organisation. Assessment can contribute to the identification of

these employees.

In order to contextualise the need for this study, the commercial airline used in the research,

utilised psychometric tests in the selection of trainees for their cadet pilot training programme.

However, the relevance and appropriateness of these psychological tests could be questioned

as part of the selection battery.

Ritson (in Muller & Schepers, 2003) postulated that the rationale for using psychometric tests in

the selection process lies in the purported ability of the testing instruments to accurately and

objectively assess an applicant’s ability to perform the work required by the job.

The importance of validation of any instruments to be used for assessment purposes is

highlighted by the ongoing developments in South African labour legislation and the implications

of the EEA in particular. According to Van Der Merwe (1999), these issues also accentuate

once again the need for the responsible use of tests and other psychological procedures.

Experience has shown that tests are generally far more reliable and more valid than other

techniques (Van der Walt, 1998). Van der Walt (1998) further stated that studies in trade and

industry have indicated that psychometric tests are about four times more effective than

5

screening interviews. The commercial airline thus exercised sound practice in using

psychological tests in the selection process for the cadet pilot training programme.

The literature on pilot selection indicates that the military introduced this process during the

World War II. The identification of candidates who are likely to succeed as military pilots has

been a longstanding goal (Carreta & Ree, 1989). According to Carretta and Ree (1996), the

selection of military pilots has always included the use of multiple aptitude test batteries. They

(1996, p. 279) further stated that “these tests have been described as measuring a variety of

abilities that have demonstrated utility in predicting pilot success”. This statement refers

specifically to two tests used by the United States Air Force (USAF) namely, the Air Force

Officer Qualifying Test (AFOQT) and the Basic Attributes Test (BAT).

During the literature review few previous research on the topic were found, a point noted by

Damon (1996) that all the studies conducted on selection batteries were concerned with fighter

pilots and no examination of concurrent validity or post-undergraduate training performance

were made for transport pilots. This observation further supports the appropriateness and

relevance of this research, which would assist commercial airlines to implement better

recruitment objectives.

It can be argued that the selection of pilots for a commercial airline should be no different from

that of the military. The job of a pilot requires a high degree of emotional stability and flying

complex aircraft on tight schedules, in all kinds of weather, requiring a great deal of ability,

extensive experience, a clear mind and coolness in emergency situations (Butcher, 2002). This

view indicates the responsibility of airline pilots towards the safety of the passengers and cargo

they carry. Furthermore, because the costs associated with the aircraft itself are immense, to

ensure the safety not only of the passengers and cargo but also the investments in aircraft,

airlines need to select and appoint people who are reliable and psychologically sound (Butcher,

2002).

In relation to the military, Flotman (2002) suggested that because of the shortage of pilots in the

South African Air Force and the acquisition of new aircraft, it is critical to enhance the value of

these newly acquired aircraft by selecting and training the most suitable personnel.

6

The above views explain the rationale for the need to investigate the utilisation of psychological

assessments in predicting successful completion of a cadet pilot training programme.

1.2 PROBLEM STATEMENT

Traditionally, the function of psychological tests has been to measure differences between

individuals or between the reactions of the same individual on different occasions (Anastasi,

1990). Psychological tests can be of great value in employee selection because of their

objectivity and validity.

The diverse nature of the South African community poses a challenge when considering the use

of psychological tests during the selection process of cadet pilots for a commercial airline.

Research investigating the predictive validity of psychological tests in pilot selection has

generally mainly been applied in the military environment. According to Hunter and Burke

(1994), military pilot selection has traditionally been heavily researched, partly because pilots

play a key role in modern warfare, and training them is costly in terms of both finances and time.

Safety is paramount in aviation. According to Hoole and Vermeulen (2003), the human factor is

widely recognised to be critical to aviation safety and effectiveness. They (2003) reported that

the numerous studies indicating that the human factor is vital in maintaining or improving safety,

prompted the development and publication in 1990 of the United States National Plan for

Aviation Human Factors for research, which would lead to the enhancement of

(1) human-centred design of controls, display and advanced systems

(2) selection and training

(3) information transfer

(4) personal safety, well-being and survival

(5) the measurement of performance and understanding of variables that affect

performance (Federal Aviation Administration, in Hoole et.al. 2003).

This research supports objective (2) above, and will contribute to the field of industrial and

organisational psychology.

The problem faced by the commercial airline is that the current selection test battery has not

been scientifically investigated for its predictive validity in the selection of “cadet pilots” for the

7

training programme. Since training is a precursor to job performance, training criteria are vital

because they impart job knowledge that facilitates the performance of the job task (Carretta &

Ree, 1996).

According to Schmidt and Hunter (1998, p. 262), “the most important property of a personnel

assessment method is predictive validity: the ability to predict future job performance, job-

related learning, such as amount of learning in training and development programs, and other

criteria”. This statement highlights the focus of this research.

It is anticipated that the current research will provide information to assist the commercial airline

to select candidates with the potential to successfully complete the training programme, thus

reducing the financial burden on the organisation.

In this research, the measurement of learning potential was not part of the psychological battery

used in the selection of cadet pilots.

The next section deals with the research aims.

1.3 RESEARCH AIMS

1.3.1 General aim

The general aim of this research is to ascertain whether the psychological tests used for the

selection of cadet pilot training programme in a commercial airline are valid predictors of their

performance using their final flying school results as a criterion measure. According to Muller

and Schepers (2003), this is in line with the guidelines of the Society of Industrial Psychology

which state that the onus is on practitioners to validate the tests they use, and to provide

concrete empirical evidence that their selection practices are fair (Society for Industrial

Psychology, 1992).

The aim of this research study is not to determine the reliability and validity of the cadet pilot

training programme.

8

1.3.2 Specific aims of the literature review

(1) To review the literature on the concepts of intelligence and aptitude and explore the

nature of these concepts, as well as how intelligence and aptitude can be measured

using psychological instruments. The concept of selection will also be covered.

(2) To review the literature on the concepts of literacy and numeracy and describe their

dimensions.

1.3.3 Specific aims for the empirical study

(1) To evaluate the predictive validity of the psychological test battery used in the selection

of cadet pilots as a predictor of success in the training programme.

(2) To compare different race groups, in terms of psychological test battery scores, in order

to investigate whether the English Matric symbol and ABET levels correlate with

performance in the final flying school examination results.

1.4 THE PARADIGM PERSPECTIVE OF THE RESEARCH

According to Mouton and Marais (1990), normal science may be defined as the practice of

scientific research within and from the frame of reference supplied by a dominant paradigm. In

this sense, a paradigm is primarily a model for conducting normal research.

The relevant paradigms, meta-context and disciplinary and methodological dimensions of this

research are discussed below and provide a context for the research within the discourse of

science.

1.4.1 Applicable psychological paradigms

Morgan (1980) stated that one of the principal implications of Kuhn’s work stems from the

identification of paradigms as alternative realities. The term “paradigm” is used in its meta-

theoretical or philosophical sense to denote an implicit or explicit view of reality. The research is

9

conducted within the broader context of paradigms and disciplines of cognitive behaviourism,

functionalism and empiricism. The literature survey on literacy and numeracy and psychological

tests in a selection process will be presented from the cognitive behaviourism, functionalism and

empiricism paradigms.

1.4.1.1 Cognitive behaviourism

According to Bergh and Theron (2003), behaviourism is based on Watson’s belief that the

subject matter of psychology should be observable behaviour, because Watson maintains that

only what is observable can be studied objectively. These authors (2003) further stated that

after 1960, social learning theories were developed which focused on learning in social

situations because of the introduction of the notion of consciousness into behaviourism theories.

This occurs through the use of cognitive processes, that is, processes whereby the individual

becomes conscious of the environment, such as in the recognition and perception of others’

behaviour.

It would appear that the behaviourist views humans as machines in terms of stimuli and

response, whereas the cognitive approach sees humans as complex machines in terms of

inputs and outputs. According to Leahey (cited in Bergh & Theron, 2003, p. 11), the association

between stimuli and response is replaced with the terms “computational state and internal

computations”.

In terms of this research, this implies that the stimuli are processed not purely in a clinical

behaviourist paradigm manner, that is, stimuli – response, but in a stimuli – organism –

response (S-O-R) approach. This indicates that the organism utilises various cognitive abilities

to make sense of the environment by processing the stimuli. For example, literacy and

numeracy are variables which may influence the response of the individual (organism) to the

stimulus of psychological assessment.

1.4.1.2 Functionalism

Functionalism was developed in the USA as a reaction to structurism. According to Bergh and

Theron (2003), psychology was seen as a practical science with its subject matter being the

functions of the mind instead of the structure or content of the mind. “The focus was on the

10

mind, as it is functional to the individual’s adaptation to the environment” (Bergh & Theron,

2003, p. 6). With regard to the current research, the functionalist approach in the literature

review in terms of literacy and numeracy is adopted in order to explain the functions that may

enable cadet pilots to successfully adapt to their environment.

1.4.1.3 Empiricism

Morgan (1980) stated that because human behaviour is assumed to be measurable, it can be

explained using statistical analysis. In other words, the research design is based on scientific

methods.

Empirical refers to tested knowledge and conclusions that are based on direct, sometimes

indirect, but systematic, repeated and incontrovertible observations and experience (Bergh &

Theron, 2003). This implies that the research can be verified to ensure certainty and extension

of knowledge in terms of the reported findings.

In using these paradigms for this research, the researcher was afforded an opportunity to study

and understand how cadet pilots assimilate the knowledge required to become a pilot. In the

literature review, the researcher considered what is required of the cadet pilot to acquire the

relevant skills and abilities and the impact of literacy and numeracy on these skills. Furthermore,

the researcher determined how these skills/abilities can be measured (i.e. the inductive phase).

Based on the hypotheses formulated, the extent to which the psychological tests that measure

these abilities, predict successful completion of the training programme was evaluated (i.e. the

deductive phase). Through empirical study, these hypotheses were statistically tested and

objectively verifiedy (i.e. the verification phase).

1.4.2 Meta-theoretical statements

According to Bergh and Theron (2003), an eclectic approach involves meta-theories, which are

integrative approaches that overcome the limitations of adhering to one particular theoretical

point of view. They further stated that the aim of metapsychology is to place human behaviour

and experience in a holistic perspective. In this perspective, the assumption is that a human is

an organised whole, functioning in totality through the interaction of structures and processes.

The intellectual climate refers to the variety of meta-theoretical values or beliefs and

11

assumptions, which because of their origin, can usually be traced to nonscientific contexts and

have become part and parcel of the intellectual climate of a particular discipline in the social

sciences (Mouton & Marais, 1990). The meta-theoretical assumptions underpinning theories

and models provide a vital contextual perspective of the research.

According to Theron (2003, p. 11), “the aim of metapsychology is to place the human behavior

and experience in a holistic perspective”. This not only studies humans in relation to their

physical environment, but also in relation to their transcendental environment, which includes

theology and philosophy (Meyer et al., in Theron, 2003). Meta-theoretical statements, in terms

of this research are presented below.

A theoretical statement refers to those beliefs of which testable statements about social

phenomenon are made. Theoretical belief may therefore be regarded as assertions about the

what (descriptive) and why (interpretative) aspects of human behaviour (Mouton & Marais,

1990). In terms of this research, this could mean that candidates, who are selected for the

training programme on the basis of their psychological tests results, should be successful in the

training programme.

A methodological statement refers to beliefs concerning the nature of social science and

scientific research. The primary methodological models are quantitative and qualitative (Mouton

& Marais, 1990).

The methodology used in this empirical study is quantitative, which is congruent with the

methodological conventions of the subdiscipline of psychometrics. The data on the independent

and dependent variables are quantitative.

The interrelations of disciplines/field of industrial psychology are as follows:

(1) Personnel psychology. According to Bergh and Theron (2003), this field is concerned with

recruitment, selection, placement and training of employees, as well as a study of factors

that affect the utilisation of personnel (also known as human resources). In this research,

the selection of personnel will be studied.

12

(2) Career psychology. This field is concerned with career and organisational choice, career

issues that affect individuals in the course of their careers and changes in the organisations

that affect careers (Bergh & Theron, 2003). In this research, choosing a career as a pilot will

be discussed.

(3) Psychological assessment. Psychological assessment is a core discipline in most fields of

industrial psychology. The focus is on studying the principles and techniques for the

assessment of individual differences and similarities within or between people (Bergh &

Theron, 2003). The use of psychological tests in predicting performance for the various race

groups in the pilot training programme will be a focus of this research

(4) Psychometrics. According to Moerdyk (2009), psychometric tests have been designed to

measure one or a small number of dimensions, and the points for the answers given can be

added together to make a single score. In psychometrics, however, there has been a

predominant interest in the quantitative study of abilities, educational achievements,

personality styles, attitude and other human traits and characteristics (Fillmore, Kempler &

Wang, 1979).

In the current research, the predictive validity of the psychological tests used in a selection

process will be discussed.

1.4.3 Market of intellectual resources

Mouton and Marais (1990) distinguished between theoretical and methodological beliefs.

The main thesis of Mouton and Marais’s model is that in a research project, the researcher

internalises specific inputs from the paradigm(s) to which he/she subscribes in a selective

manner, to enable him/her to interact with the research domain in a fruitful manner and to

produce scientifically valid research (Mouton & Marais, 1990).

According to Mouton and Marais (1990), the principle of selective internalising may explain why

researchers do not necessarily adhere to an identifiable paradigm in their research. This implies

that the researcher in the current research and in the literature review will consider whether the

levels of literacy, (particularly English literacy) and numeracy, influence performance in

psychological tests for predicting success in the cadet training programme.

13

1.4.4 Psychological models and theories

Cattell’s (1987) theory of fluid and crystallised intelligence underscores most of the literature

review. However, other models and theories on intelligence and cognitive abilities are also

important for the research. Some of these are Spearman’s (1904) two-factor theory,

Gustafsson’s (1989) three-level model, Thurstone’s (1938) primary mental abilities and Carroll’s

(1993) three-stratum model.

1.4.5 Applicable concepts and constructs

The following concepts and constructs are applicable to this study:

(1) Intelligence. Moerdyk (2009) states that according to the About Intelligence newsletter,

intelligence is often seen in the popular sense as the general mental ability to learn and apply

knowledge to manipulate one’s environment, as well as the ability to reason and show abstract

thought. More formally, intelligence involves the ability to act purposefully by seeking out

relevant knowledge, finding or creating rules to link these units of knowledge or information,

applying them to novel situations, adapting and extending the rules where necessary, and

generally arriving at a reasonable and appropriate decision (Moerdyk, 2009).

According to Bartholomew (2004), the problem with the concept is that it is twisted so that it can

be empirically tested, and intelligence is not about doing tests, which tend to be concerned not

only with speed of performance and suchlike, but also about creativity and the willingness to

grapple with ideas. Measurement of cognition concentrates on problem solving and information

processing (Bornstein, Schuster & Ashburgn, 1992).

(2) Literacy. According to Holme (2004), viewing literacy as the mastery of a set of skills can

mean that this mastery is in fact never attained. The indeterminate nature of what being literate

means was clearly acknowledged in a UNESCO report: literacy is a characteristic acquired by

individuals in varying degrees from just above none to an indeterminate upper level. Some

individuals are more or less literate than others, but it is really not possible to speak of illiterate

and literate people as two distinct categories (Holme, 2004).

14

(3) Functional literacy. In his discussion of literacy, Holme (2004)commented that the problem of

literacy measurement is intensified in a developed economy where schooling, which

sometimes resulting in educational failures, is at least universal. In the author’s view there

are not two contrary states, literate and illiterate. Functional literacy recognises the fact that

people will master the skills of literacy to different degrees. Functional illiteracy occurs when

the ability to read and write is inadequate for an individual to engage in society, work

effectively and pursue lifestyle choices (Holme, 2004).

According to Ellsworth, Hedley and Baratta (1994), functional literacy is the capacity to use

language to do something, say, read an advertisement or uses a technical manual to fix a

leaking sink. They further stated that critical literacy includes the capacity for action, but also

incorporates a broader sense of understanding and insight and the ability to communicate

with others about “texts” that are either written or spoken.

(4) Numeracy. According to Harris and Hodges (1995), numeracy is explained as fluency in

mathematical operations. Bergh (2009) described numerical ability as the ability to reason

quickly and accurately by way of addition, subtraction, multiplication and division.

(5) Validity. This is the psychometric requirement for a measurement technique to measure

the construct it is designed to measure (Bergh, 2003).

Several categorisation systems are used, but in this research, the following were

considered:

According to Moerdyk (2009), there are three main forms of validity. They are all important,

but are applied differently in different contexts and therefore require different kinds of

evidence. The three forms of validity are as follows:

Construct (theoretical) validity. This is concerned with whether the assessment

technique produces results that are in line with what is already known (Moerdyk, 2009).

Content validity. According to Moerdyk (2009), content validity is concerned with whether

the content of the scale or measure accurately reflects the domain it is trying to assess.

This form of checking is most appropriate for achievement and knowledge assessment.

15

Criterion-related (empirical) validity. This relates the scale outcomes to some external

criterion. Moerdyk (2009) describes the following two forms of criterion-related validity:

concurrent and predictive validity:

- Concurrent validity. This form of validity is designed to ask whether the measure

successfully distinguishes between known groups.

- Predictive validity. This form of validity is focused on whether the assessment

procedure can predict how groups may differ in the future (Moerdyk, 2009).

There are two further forms of validity worth mentioning here, namely face and ecological

validity.

Face validity. According to Moerdyk (2009), the basic issue with face validity is that the

assessment technique should appear (especially to the uninformed) to be doing what it

claims to be doing.

Ecological validity. This is concerned with whether the results of the assessment are

meaningful and useful outside the setting in which they are obtained.

(6) Reliability. According to Bergh and Theron (2003), reliability involves the consistency of

measurements, that is, a process or measurement repeated in various situations or by

different people will provide more or less the same measurement results.

In this research, the predictive validity of the psychological tests used during the selection

process of the “cadet” pilots to successfully complete the pilot training programme was

evaluated using the final flying school results as the criterion.

1.4.6 Methodological assumptions

Scientific research is a systematic, controlled, empirical and critical investigation of material

phenomena guided by theory and hypotheses about the presumed relationships between such

phenomena (Kerlinger, 1986). This research is quantitative, in accordance with scientific

methods of research, and is aligned to the subdiscipline of psychometrics. When processing

16

data relating to the dependent and the independent variables, statistical formulae will be used to

provide quantitative results which may indicate whether or not there is a relationship between

the dependent and the independent variables.

1.5 RESEARCH DESIGN

The aim of a research design is to plan and structure a given research project in such a way

that the ultimate external and internal validity of the research findings is maximised (Mouton &

Marais, 1990).

According to Cooper and Schindler (1998), the following are the essentials of research design:

The design is a plan for selecting the sources and types of information used to answer

the research question.

It is a framework for specifying the relationships between the study’s variables.

It is a blueprint that outlines each procedure from the hypotheses to the analysis of data.

In the current research, the data were collected by means of psychometric and academic

assessment, which is a nonexperimental correlational design.

1.5.1 Research variables

According to Kerlinger (1986), research design sets up the framework for studying the

relationships between variables. The independent and dependent variables in this research are

discussed below.

1.5.1.1 Independent variables

The independent variables are the scores for each individual in the battery of the psychological

tests in phases 1 and 2, namely: English Literacy Skills Assessment (ELSA); the Blox Test;

Raven’s Progressive Matrices (RMP); the Intermediate Test Battery; and the Wechsler Adult

Intelligence Scale (WAIS).

17

1.5.1.2 Dependent variables

The dependent variables are the results obtained in the final examination at the flying school

offering the cadet pilot training programme.

The research is descriptive in both the review of the literature and the empirical study..

According to Mouton and Marais (1992), the goal of the researcher in this type of research is to

describe that which exists as accurately as possible. The above-mentioned will be integrated in

the conclusion, and recommendations will be contextualised with reference to the formulated

problem.

1.6 RESEARCH METHOD

This research was undertaken in two phases, namely a literature survey and an empirical study

which are outlined below.

1.6.1 Literature review

The specific aim of the literature review is to conceptualise intelligence, literacy and numeracy.

These concepts are defined, focusing on the nature of the concept of intelligence and how it can

be measured by means of psychological tests.

1.6.1.1 The use of psychological tests in the selection of cadet pilots

In chapter 2, the concepts of psychological test, intelligence and selection are defined and

discussed. The theories of intelligence are explained in relation to the concept.

1.6.1.2 The nature of literacy and numeracy

The literature review is focused on the conceptualisation of literacy and numeracy, the nature of

these concepts and how they are operationalised. In chapter 3, literacy and numeracy are

defined to provide a conceptual understanding of the domain.

18

1.6.1.3 Integration

The aim of this section of the literature survey is to integrate available information on literacy

and numeracy and psychological tests in the selection process in order to demonstrate the

theories and models that are dealt with in chapters 2 and 3 and which are used as the basis to

test the relationship empirically.

1.6.2 Empirical study

The empirical research is summarised and the population and sample described. The

psychological instruments used in the study are identified and the selection of the psychological

instruments highlighted. The manner in which data were collected and processed is discussed

as well as process of testing the hypotheses. A general outline of how the results will be

reported and interpreted is given and lastly, conclusions are drawn, the limitations of the study

discussed and recommendations made (see aim 1.3).

1.6.2.1 Data collection

All testing was done at the regional recruitment centres. The administration and scoring were

done by a registered psychometrist under standard testing conditions. The test results and

scores were checked and captured by an administrative assistant. Chapter 4 explains the data

collection process.

1.6.2.2 Data processing

In order to determine predictive validity, the Pearson correlation coefficient and regression

analysis were used to measure the relationship between the independent (predictor) and

dependent (criterion) variables. Chapter 4 discusses the processing of the data that were

collected.

1.6.2.3 Research hypotheses

The research hypotheses were formulated with regard to the primary aim of the study in order to

determine whether there is a statistically significant relationship between the psychological test

19

scores and the final flying examination results. Regarding the second aim of the study, the

hypothesis states that there is statistically significant relationship between scores of the final

examination flying school results and English Matric symbol and ABET levels. These

hypotheses will be discussed in chapter 4.

1.6.2.4 Reporting of results

Chapter 5 reports on the predictive validity of the empirical results and of the psychological tests

used in line with the specified aims of the research, as outlined in the empirical study.

1.6.2.5 Conclusions, limitations and recommendations

Conclusions are drawn on the basis of the extent to which the research aims have been met.

Limited resources in terms of human capital and time and the follow-up of successful applicants

who joined the organisation influenced the limitations of this research. Further limitations in

terms of statistical considerations will be discussed in chapter 6.

Chapter 6 also makes recommendations for further research. These include follow-up studies in

terms of job performance as predicted by the psychological instruments. Furthermore, the

standardisation of the psychological test batteries aligned to competencies required for a

competent pilot is proposed.

1.7 CHAPTER LAYOUT

Chapter 2: The use of psychological tests in selection and nature of intelligence

Chapter 3: The concepts of literacy and numeracy

Chapter 4: The empirical study

Chapter 5 The research results

Chapter 6 Conclusions, limitations and recommendations

20

1.8 CHAPTER SUMMARY

In this chapter, the background to and rationale for this research were discussed. The problem

statement, the aims, the research model, the paradigm perspective, the research design and

method and the chapter layout were also highlighted.

21

CHAPTER 2

THE USE OF PSYCHOLOGICAL TESTS IN SELECTION WITH SPECIFIC FOCUS ON

MEASUREMENT OF COGNITIVE ASPECTS

2.1 INTRODUCTION

The notion of human assessment as a means of knowing and understanding another person

has been around for a long time. Human behaviour has been assessed in a variety of

settings and by individuals in many different disciplines (Walsh & Betz, 1995).

The recruitment of people as pilots in the aviation industry, in a multicultural society, poses a

challenge, particularly in South Africa. Owing to the huge costs associated with pilot training,

selecting candidates to successfully complete the pilot training programme has resulted in a

special focus on the selection process.

In line with the objective of this research, namely to determine the predictive validity of the

psychological tests to select cadet pilots, this chapter will discuss the selection of pilots and

the psychological tests included in the test battery. One of the aims of the current research is

to investigate variables that have positive correlations with the successful completion of the

cadet training programme. According to Carretta (1992), medical and physical fitness, paper-

and-pencil aptitude test scores, academic performance and previous flying experience are

variables currently considered in pilot candidate selection in the USAF.

The concept of selection will be explained in relation to the use of the psychological tests

utilised during the process of pilot candidate selection. The various predictors, namely

intelligence/aptitude, English Matric symbols and the ABET levels will also be discussed.

2.2 DEFINITION OF CONCEPTS

Bartram (2004) commented that in the workplace, tests, as part of assessment, are used to

measure the performance and potential of current and future employees through selection

22

and performance management respectively. This suggests that psychological assessments

in organisations need to be used responsibly, ethically and equitably.

Moerdyk (2009) argued that in South Africa, with the socioeconomic differences continuing

to be visible in the education and general socioeconomic status of different sectors of the

country‟s population, large parts of society remain at a disadvantage when they are

assessed with current assessment techniques and tools such as psychological tests. This

implies that the use of psychological tests in their adapted and/or current format may unfairly

discriminate against people, particularly in a multicultural society such as that of South

Africa.

2.2.1 Selection process

In most organisations, recruitment and selection are the responsibility of human resources,

who are tasked with recruiting and selecting suitable candidates to perform duties on the

basis of their abilities. Unfortunately, there has been and still is a trend towards a large

number of people with the abilities and/or competencies to perform a specific job and/or

occupy a certain position in an organisation applying for very few jobs and positions. After

attracting as many suitable applicants as possible, the process of selection commences.

Selection methods vary between two extremes on a continuum, from ”scientific approaches”

at one extreme, to an ”intuitive approach”, at the other (Louw & Edwards, 1997). According

to Louw and Edwards (1997), the intuitive approach is more subjective, whereas in the

scientific approach, efforts are made to limit subjectivity. The requirements of the EEA

support the scientific approach in the selection process.

According to Damos (1996), the term “selection battery” refers to either batteries

administered as part of screening for air carriers that hire experienced pilots or to batteries

administered to ab initio pilots, prior to admission to flight training or during the ground phase

of training. In the current research, the focus is on ab initio pilots.

Literature on pilot selection suggests that much was achieved in the military institutions

(Damos, 1996; Martinussen & Torjussen, 1998; Ree & Carretta, 1996; Tsang & Vidulich,

2003), because commercial airlines used to recruit their pilots from the military. According to

23

Hunter and Burke (1994), the bulk of published reports have dealt with selection for ab initio

(beginner) pilot military training, although much of what has been accomplished in the

military setting is directly applicable to civic aviation.

In a study by O‟Hare and Walsh (2003) which involved a review of articles published during

1996 and 2000, with particular focus on the International Journal of Aviation Psychology

(IJAP), which was the first journal devoted exclusively to aviation psychology, they

established that of the 21 topics discussed in IJAP, 21% of the articles dealt with display

issues, 13% with training, 10% with automation and 9% with selection. This observation led

the authors to conclude that there is growing interest in the articles on display and

automation and a decreasing proportion on workload. Hence there has been no sign of a

broadening of the content base beyond the traditional flight deck and ATC concerns. This is

not surprising in the light of the critical issue of flight safety in the airline business.

Locally, the literature review shows trend that is similar to the international one, from a

research perspective. However, the research conducted locally was in the military and

extremely limited in commercial and/or civil aviation. De Kock and Schlechter (2009) shared

this view in their statement that traditionally, validation research received considerable

attention in the military.

Military pilot selection has been heavily researched for a number of reasons.

For instance, the literature indicates that training attrition rates over the last 20 years have

typically been in the order of 25%, with an average cost for each failure ranging from $ $50

000 to 80,000 for the USAF (Hunter & Burke, 1994; Martinussen, 1996). De Kock and

Schlechter (2009) concurred when they stated that in the UK, the estimated unit cost of

training a fast-jet pilot is more than £3,7 million. In the South African Air Force (SAAF), it

takes five years to train a fighter pilot.

These views confirm that training failures are costly. Having an efficient and properly

structured selection process should ensure that there is a better “fit” between the selected

applicants and the job, which would also ensure that the organisation obtains the “best”

return on its investment. Also confirmed in relation to selection, is that it is essential to admit

people to a training programme, with a high probability of success, thus minimising a

mismatch with the position and reducing employment costs.

24

According to Van Der Merwe (2002), the vast majority of employee selection programmes

are based upon the successive hurdle technique. This means that in order to be hired, the

applicant must successfully pass various screening steps. At each step or hurdle, some

candidates are rejected. The selection process, as indicated by Van Der Merwe (2000), in

figure 2.1, shows the method used during cadet pilot selection in the commercial airline.

FIGURE 2.1. The selection process Source: Adapted from Van Der Merwe (2002, p .78)

The process is aligned to those in the military, as noted in Carretta and Ree (2003), who

stated that selection for the military or civilian pilot training programme is typically a

multistage process, in which decisions are made at several points.

One of the critical elements in ensuring outstanding organisational performance is the

selection and development of excellent staff. International as well as local research has

demonstrated the role that psychometric assessment can play in significantly improving the

Recruitment of candidates

Application forms

Preliminary screening

Reference checking

Psychological testing

Interview: line management

Medical examination

Successful candidates Unsuccessful candidates

Rejectio

n

25

selection process for both new entrants and internal promotions (Van der Merwe, 2002).

Effective psychometric assessment can also play a key role in staff development processes,

and this is one of the main challenges presently facing South Africa (Van der Merwe, 2002).

It can be inferred that selection is a vital process in identifying candidates with the potential

to enter and/or join an organisation and to reach their optimum performance with relevant

and appropriate training and development.

2.2.2 Definition of selection

The general objective of a selection process is to match the person to the job. As noted in

the above discussion, the selection of a pilot for training is not that different from a selection

process in any organisation. The term “selection” will be broadly defined and common

features of a selection process highlighted.

Moerdyk (2009) defined selection as the process of matching people to the job requirements

in order to meet organisational objectives, both current and in the longer term.

According to Jackson (1996), who described selection from a decision-making perspective,

the concept involves only two categories – acceptance and rejection.

According to Louw and Edwards (1997), some of the features common to most selection

processes are as follows:

Through the job analysis process, the traits (or characteristics) required to perform

the job competently are identified.

Information is obtained about predictors.

An evaluation is made of how well the predictors succeed in predicting the

performance of applicants.

These features indicate that there should be close fit between the applicant and the job and

that decision making should be based on objective information obtained during the selection

process using psychological tests.

In their definition of pilot selection, Carretta and Ree (2003) asserted that in military aviation,

the goal is to achieve and maintain a high level of mission readiness – hence the fact that

26

organisations need people to serve in various capacities and people need jobs. They further

stated that to achieve this, the needs of the organisation should be matched to those of the

job applicant. Making the right selection decision thus reduces training costs, improves job

performance and enhances organisational effectiveness (Carretta & Ree, 2003).

Effective selection procedures will produce cost avoidance savings through reduced attrition

and reduced training requirements and will improve job performance. Poor selection will

have the opposite effect on the organisation (Carretta & Ree, 2003). Moerdyk (2000)

expressed a similar view when he noted that poor selection can result in quality and safety

being compromised, increased risk of injury, underspending of allocated budgets and

underdelivery of vital services, to mention but a few, in addition to direct costs. The

implications of these views indicate that the commercial airline should aim to have a

streamlined selection process that should impact positively on service delivery because of

the type of people appointed in the various positions and capacities.

It is common practice for most organisations to use psychological tests in their selection

process as an aid to decision making. In support of this view, Elkonin, Foxcroft, Roodt and

Astbury (2005) stated that perhaps the mostly widely used function of psychological

measures in industry is to assist with selection and employment decisions. Anastasi and

Urbina (1997) concurred with this view when they stated that the goal of measuring human

attributes in a work situation is to identify the potential of individuals and to match them with

the right job.

2.3 PSYCHOLOGICAL TESTS

It is becoming increasingly common for psychological tests or entire test batteries to be used

in the selection of suitable candidates for a professional position or training programme,

although internationally, the frequency of their use varies greatly (Schuler, in Sommer,

Olbrich & Arendasy, 2004). Sommer et al. (2004) noted that the derivation of decisions and

training from the results of such individual tests requires a sufficiently high level of

agreement between the test used and the corresponding criterion variable.

27

According to Nzama et al. (2008), in South Africa in particular, the use of psychological

assessment as a means of determining the employability of individuals has had a mixed

history, marked by acceptance, in some instances, and by scepticism, in others.

Aligned to this view is the comment on ability tests by Muller and Schepers (2003), who

reported that Wood and Payne (1998) found that the proportion of organisations using tests

to select staff rose from just below 50% in 1991 to 75% in 1996, making them as popular as

curriculum vitaes.

It is against this background that the ensuing discussion will consider the definition and

functions of psychological tests in a selection process.

2.3.1 Definition of psychological tests

Throughout the discussion thus far, psychological tests have been generally defined as an

objective and standardised measure of a sample of behaviour. Psychological tests are like

tests in any other science, insofar as they are observations on a small but carefully chosen

sample of an individual‟s behaviour (Anastasi, 1990). This definition indicates that, if a

psychologist wishes to measure a clerk's ability to perform arithmetic computation or a pilot's

hand-eye coordination, he/she must examine their performance by means of a

representative set of arithmetic problems or motor tests (Anastasi, 1990).

Cohen and Swerdlink (2002) defined psychological assessment as the gathering and

integration of psychology-related data for the purpose of making a psychological evaluation,

accomplished through the use of tools such as tests, interviews, case studies and

behavioural observation, as well as specially designed apparatuses and measurement

procedures. They define psychological testing as the process of measuring psychology-

related variables by means of devices or procedures designed to obtain a sample of

behaviour.

According to Shum, O‟Gorman and Myors (2006), a psychological test is an objective

procedure for sampling and quantifying human behaviour to make inferences about a

particular psychological construct using standardised stimuli and a method of administration

and scoring.

28

The assessment process is multidimensional and entails gathering and synthesising

information as a means of describing and understanding functioning (Foxcroft & Roodt,

2005).

According to Elkonin et al. (2005), basically, two approaches are used in the application of

psychological measures. In the input-based approach, individuals are compared with the job

specifications in terms of their personal characteristics or personality traits. The other

approach is output based in the sense that individuals are compared in relation to the

required outputs of the job. In this instance, the aim is to determine whether the individual

has the necessary competencies to perform a particular task or job.

In terms of this research, the output based approach is used to determine whether the

applicants have the ability to successfully complete the cadet pilot training programme.

Based on the above definitions, one may infer that psychological tests may differ according

to the number of investigated variables such as content, format, administration procedures,

scoring and interpretation procedures and psychometric or technical quality.

These views suggests that the next step in the discussion is to understand the functions of

psychological test in order to use the information obtained during assessment for the benefit

of both the applicants and the organisation.

2.3.2 Functions of psychological tests

According to Anastasi (1990), although intelligence tests were originally designed to sample

a wide variety of functions in order to estimate the individual's general intellectual level, it

soon became apparent that such tests were fairly limited in their coverage.

Aiken (1991) emphasised this view when he stated that because employment tests, which

are classified as psychological tests, had been validated primarily on members of the

dominant white culture, it was reasonable to ask whether they would be valid for black and

other minorities.

29

Cascio (1991) claimed that if an individual from a specific population group does not have an

equal opportunity of being selected for a specific post, but has an equal probability of

succeeding in the job, test bias exists which could result in unfair discrimination. Although

the current study will not be investigating the fairness of the instruments, it is necessary to

note these comments because they may impact on the commercial airline not achieving its

employment equity targets, as a requirement of the EEA.

According to Anastasi (1990), one needs to be sure that the tests cover abilities that are of

primary importance in one‟s culture.

Muller and Schepers (2003) asserted that it is important to acknowledge the fact that a test is

designed to discriminate between candidates with higher and lower abilities on certain

criteria. The issue then, according to these authors, is whether the test discriminates fairly.

The main purpose of psychological testing today is the same as it was throughout the 20th

century, namely to evaluate behaviour, cognitive abilities, personality traits and other

individual and group characteristics in order to assist in making judgements, predictions and

decisions about people (Aiken, 2003). More specifically, tests are used to

screen applicants for jobs in the education and employment context

counsel and guide individuals for educational, vocational and personal purposes

retain or dismiss, promote and rotate students or employees in educational and

training programmes and in on-the-job situations

diagnose and prescribe psychological and physical treatments in clinics and hospitals

evaluate cognitive, intrapersonal and interpersonal changes caused by educational

psychotherapeutic and other intervention programmes

conduct research on changes in behaviour over time and evaluate the effectiveness

of new programmes or techniques

For the purpose of this research, the psychological tests were utilised to evaluate the

cognitive abilities of applicants for the cadet pilot training programme and to determine

whether or not they could successfully complete the training programme.

This view was emphasised by Johnston (1997, p.179) who confirmed that “psychological

testing of applicants for pilot training and employment is now a well-established and widely

30

accepted activity in most parts of the world; it is only in very recent times that psychological

testing has been potentially envisaged to have a substantive role in pilot licensing”.

Foxcroft and Roodt (2005) highlighted some of the problems encountered when changes

were evident in the sociopolitical situation in South Africa. In the first instance, measures

were developed for more than one racial group and/or norms were constructed for more than

one racial group so that test performance could be interpreted in relation to an appropriate

norm group. In the second instance, psychological measures were developed and

standardised for white South Africans only, and some were also imported from overseas and

used to assess other groups as well. These authors (2005) further stated that in the absence

of appropriate norms, the potentially bad habit arose of interpreting such test results “with

caution”. The major problem with this approach was that initially little research was done to

determine the suitability of these measures for a multicultural South African environment.

Another area of psychological testing is concerned with the affective or nonintellectual

aspects of behaviour. Tests designed for this purpose are commonly known as personality

tests, although some psychologists prefer to use the term "personality" in a broader sense to

refer to the entire individual. As stated by Anastasi (1990), in the terminology of

psychological testing, however, the designation "personality test" generally refers to

measures of such characteristics as emotional states, interpersonal relations, motivation,

interests and attitudes.

As mentioned earlier, personality tests were not part of the selection battery used during the

assessment of the applicants for the purpose of this research.

It is evident from the previous discussion that psychological tests can be used for many other

purposes such as to determine individual differences in general intelligence, specific

aptitudes and noncognitive personality traits.

Groth-Marnat (2003) suggested that intelligence tests are limited in predicting certain

aspects of occupational success and nonacademic skills such as creativity, motivational

level, social acumen and success in dealing with people. Finally, there has been an

overemphasis on understanding the end product of cognitive functioning and relative neglect

in appreciating underlying cognitive processes (Groth-Marnat, 2003).

31

Dillon (1997) raised a concern, not with the particular type of test, but with the method of

measurement. The author‟s premise is that neither type of tests will account adequately for

intelligence performance when test scores are used as predictors. Dillon (1997) proposed

that the same kinds of tests may continue to be used or new tests may be developed. In

either case, indices of ongoing information processing should be recorded, and such

measures should be used as predictors of school or occupational success.

In the current research, the applicants were assessed on their intellectual skills on the basis

of the scores obtained from the various measurement instruments. As indicated in the above

statement, the underlying cognitive processes were not assessed - in other words, the

assessment answered the question “what” and not “how”.

Van Eeden and De Beer (2005) stated that psychological measures are not as accurate as

physical measures such as height. This exacerbates the difficulties and controversy

surrounding the measurement in social sciences, which is partly due to the fact that one

measures constructs which are entities not directly visible or directly measurable and which

have to be measured in a roundabout fashion. Hence measuring variables such as

intelligence and aptitude will have to be inferred from the scores obtained by the candidates.

Ree and Carretta (1996) reported on research by Yerkes (1919) during World War I, which

showed that measures of intelligence were valid predictors of pilot training success. The

authors further reported that between the two World Wars, Flanagan (1942) noted that the

US aviation selection examination involved a general mental battery testing comprehension

and reasoning – again highly “g” loaded.

According to Ree and Carretta (1996), World War II generated a renewed interest in pilot

selection by all combatants. Influenced by multiple aptitude theory (e.g. Melton, 1947) and

the Navy (e.g. Fiske, 1947; Viteles, 1945), Ree and Carretta (1996) noted that the US Army

used several ability measures for pilot selection, including intelligence, psychomotor,

mechanical comprehension and spatial measures.

These views suggest that inferences need to be made about the constructs, that is, traits

that are supposed to be measured by the measurement instruments. When selecting

32

applicants for ab initio (beginner) pilot training, indicators of ability (i.e. trainability) are

emphasised (Carretta & Ree, 2003).

Psychological tests are thus tools that are seen as objective in the assessment of individuals

in order to assist decision making in a number of different situations (Van Der Merwe, 2002).

The focus is normally on a sample of behaviour considered important in an organisation.

Psychological tests can be considered useful when they can provide information and/or data

that are perceived to be valid and reliable. Legislation, as outlined in the Skills Development

Act 97 of 1998, emphasises the fact that psychological tests may be used in the selection

process, provided their validity and reliability can be proven.

This leads to the next point of discussion in considering the determinants of pilot training

success.

2.4 THE DETERMINANTS OF PILOT TRAINING SUCCESS

According to Carretta and Ree (1996), there is consensus that the determinants of pilot

success reside in the following three main domains:

2.4.1 Intelligence and aptitude

Hilton and Dolgin (1991) have characterised intelligence as the best and most stable

predictor of flight training success, based on their summary of pilot selection research during

the last century. According to De Kock and Schlechter (2009), intelligence is a broad concept

and is sometimes defined more specifically. These authors referred to Ree and Carretta‟s

(1996) distinction between two types of intelligence, where they used Spearman‟s (1904)

two-factor theory of cognitive ability and argued that intelligence can be seen as a general

cognitive ability “g”, on the one hand, or in terms of specific abilities (Sn), on the other.

According to De Kock and Schlechter (2009), the construct “g” is synonymous with fluid

intelligence.

2.4.2 Psychomotor coordination

Measures of psychomotor coordination or hand-eye coordination, as it sometimes referred

to, are commonly included in the selection batteries for two apparent reasons, namely (1)

33

that they have an obvious relation to the task; and (2) the results of validation research

support their inclusion in selection batteries (Hilton & Dolgin, 1991).

2.4.3 Personality

Contrary to expectation, most studies report that personality adds little to the prediction of

pilot success (Carretta, Perry & Ree, 1996; Hunter & Burke, 1994; De Kock & Schlechter,

2009). Some studies did in fact report that certain aspects of personality had incremental

predictive validity in traditional batteries, for instance, attitude to risk (Ree & Carretta, 1996).

The focus of the current research is on one of the domains, namely intelligence and aptitude,

as determinants in predicting success in a pilot training programme for the commercial

airline. The data for the other two domains were not available because they had not been

collected or captured during the assessment. Also, the limited nature of the research had an

impact.

2.5 THE DEVELOPMENT AND THEORIES OF INTELLIGENCE

The development of intelligence is discussed by examining the theories of the concept of

intelligence as part of cognitive ability. This will be followed by a look at the measurements of

the concept.

As in any scientific practice, theory formulation plays a cardinal role, and the aim of any

theory is to explain logically and in simple terms the relationships between as many

perceptible phenomena as possible.

The literature on intelligence indicates that the majority of theories and models on the nature

of intelligence are based on Spearman‟s (1904) theory of “g” as a point of departure.

The theories of Spearman (1904), Thurstone (1938), and Cattell (1987) are discussed as

concepts of general “g‟ and specific “s” abilities. The works such as those of Vernon (1969)

on the theory of the hierarchical model, Guildford on the intellectual model and Jensen‟s

(1986) on the mental functioning model place the emphasis on “g” (the general factor) and

“s” (special abilities) as being inseparable.

34

These views seem to suggest that intelligence and aptitude are in fact “two sides of the

same coin”.

Spearman‟s two-factor model of general intelligence will be discussed first because it seems

to provide the foundation for most theories of intelligence. Thurstone‟s primary mental

abilities model appears to challenge Spearman‟s theory, by suggesting that intelligence is a

unitary ability and that it has a number of what he referred to as primary mental abilities

(PMAs). Building on Thurstone‟s PMAs, in his model, Cattell suggests that general

intelligence can be split into what is termed “fluid and crystallised intelligence”. Gustafsson‟s

three-level model incorporates Spearman‟s two-factor model, Thurston‟s PMAs and Cattell‟s

models of intelligence, and last but not least, Carroll‟s three-stratum theory.

2.5.1 Spearman’s two-factor model

Building upon the earlier work of Galton and Pearson, the inventors of the correlation

coefficient, Spearman (1904) concluded that all the tests he conducted measured a common

factor. This common factor is “g” – general intelligence. All the tests did not produce exactly

the same results; each had an associated smaller, more “specific” ability, which he named

“s” – specific abilities. Spearman also developed factor analysis, which determines whether

some pattern occurs in the test scores. According to Thurstone (1938), the criticism of the

two-factor model poses the question: Is intelligence a unitary ability?

Jensen (1986) provided evidence that Spearman‟s “g” is more than a statistical artifact

originating from factor analysis by showing that it is correlated more highly than any other

known factor with a number of phenomena that are independent of psychometrics and factor

analysis.

2.5.2 Thurstone’s primary mental abilities model

Thurstone (1938) found that a number of PMAs are all basic, with no one assuming more

importance than another the PMAs are as follows:

Verbal comprehension is the ability to understand the meaning of words.

35

Word fluency is the ability to think of words rapidly, as in rhyming words and

anagrams.

Number is the ability to work with numbers and compute.

Space is the ability to visualise space, for example, recognise figures in various

orientations.

Memory is the ability to recall word-pairs/sentences.

Perceptual speed is the ability to grasp visual details quickly.

Reasoning is the ability to find general rules on the basis of partial information.

Thurstone (1938) felt that a broad profile of individual mental abilities is more useful than a

simple overall measure.

2.5.3 Cattell’s model

Cattell (1987) argued that Spearman‟s theory was clearly a two-factor theory, that is, every

measured ability is a mixture of general intelligence and some quite specific ability peculiar

to that field. Thus, although verbal and numerical ability are highly loaded with Spearman‟s

“g”, each has an additional something of a special primary ability in it. Cattell viewed these

two factors, that is, fluid and crystallised intelligence, as distinct but correlated (Aiken, 1991).

While this is consistent with the spirit and technical methods of Spearman‟s basic approach,

it leads one to conclude that one is in fact dealing with two broad or “general” ability factors,

namely fluid and crystallised intelligence (Cattell, 1987).

According to Cattell (1987), one of the most interesting features is that although crystallised

ability generally does not enter culture fair subtest performance, fluid performance does,

although to a lesser extent, into those primaries such as verbal, numerical and reasoning

abilities which have been used in traditional intelligence tests.

Crystallised intelligence or – “gc” – is a measure of the outcome of culture and educational

experience. Intelligence is more dependent on the physiological structure that supports

intellectual behaviour than on crystallised intelligence. Also, it is more sensitive to the effects

of brain injury.

36

Crystallised intelligence, which reflects cultural assimilation, is highly influenced by formal

and informal educational factors. The argument against this model is that it is difficult to split

“g” into “gf” and “gc”, because “gf” and “gc” tend to cooperate (Pyle, 1979).

Cattell (1987) also argued against the concept of general intelligence because he believed

that intellectual ability is composed of several distinct functions that probably have genetic

bases. He proposed a four-level mechanical model. The usual visual and auditory sensory

detention functions are at the lowest level. The second level involves associational

processes, both short and long term. At the third level, perceptual organisational processes

come into play – broad visualisation, clerical speed and broad auditory thinking. The highest

level involves the education of relations‟ fluid ability and crystallised ability. Abilities at the

bottom of the hierarchy have low correlations with those near the top.

De Kock and Schlechter (2009) pointed out that there are two issues that seem to dominate

current pilot selection research. First, the fact that Spearman‟s general cognitive ability “g”

plays such a central role in predicting pilot success has raised the question of whether it

really makes sense to also assess specific intelligence (Carretta & Ree, 1989; Ree &

Carretta, 1996; 2002). According to these authors, the proponents of this argument

contended that because most specific intelligences are saturated with “g”, it rarely adds any

incremental validity to batteries already containing measures of “g”. This argument would

make sense, assuming Carroll‟s (1993) model of cognitive ability, which will be discussed

below, with “g” at its apex and group factors at successively lower levels to be true. For the

purpose of this research, crystallised intelligence and “g” were assumed to be theoretically

congruent.

The second issue, according to De Kock and Schlechter (2009), relates to the so-called

“criterion problems”. Most validity studies cite the lack of meaningful, quality criteria to

validate predictors against a weakness (Hunter & Burke, 1994). This problem is especially

prominent in pilot selection research (Damos, 1996).

The current research sought to address some of these issues by assessing both measures

of “g” and specific intelligence and consider how well these predict multiple criteria of pilot

training in a commercial airline.

37

2.5.4 Gustafsson’s three-level model

Gustafsson (1989) proposed a three-level model to account for the structure of intellectual

abilities as depicted in figure 2.1 below. This model integrates Spearman‟s (1904), Cattell‟s

(1971), Thurstone‟s (1938) and Guildford‟s (1967) models.

At the highest level is “g”, which represents Spearman‟s conception of intelligence. The next

level comprises three broad factors: crystallised intelligence (dealing with verbal information),

fluid intelligence (dealing with adaptive nonverbal abilities) and general visualisation (dealing

with figural information). Fluid intelligence is essentially the same as general intelligence or

“g” in this model.

Figure 2.2. Gustafsson’s hierarchical model of intelligence Source: Adapted from Sattler (1988, p. 50)

General intelligence

G

Crystallised intelligence

CG

Fluid intelligence

FG

Visualisation intelligence

VG

Verbal

achievement

V Ach

Numerical

achievement

N Ach

Verbal

comprehension

VC

Spatial

orientation

Flexibility

of closure

Visualisation

VZ

Cognition of

figural

relations

CG

Speed of

closure

CS

Induction

I

Memory

span

MS

38

Gustafsson (1989) conceived crystallised intelligence as representing a relatively narrow

dimension of knowledge and generalising less to subsequent problem-solving and learning

situations than fluid intelligence. The crystallised and fluid intelligence factors are similar to

those of Cattell and Horn‟s model of intelligence. At the lowest level are the primary factors,

similar to those in the Thurstone and Guildford traditions.

While the hierarchical model depicted in figure 2.2 above may not fit the complexities of the

interrelationships of human ability perfectly, it is a useful approximation (Humphreys et al.,. in

Sattler, 1988), and provides a useful guideline for understanding the nature of intelligence.

The view of unitary and divisible performance on literacy and numeracy tasks (to be

discussed in chapter 3) can be associated with the theories of single factor such as “g” and

separate primary mental abilities. Hence in an attempt to measure those aptitude abilities

associated with separate competence, measures of primary mental abilities may be used.

From this discussion it can be deduced that intellectual/aptitude abilities can be related to

performance in literacy and numeracy tests as broadly verbal and numerical abilities. In

other words, measures of crystallised fluid intelligence and general visualisation as depicted

in figure 2.1 may be used to measure the intellectual abilities associated with the concept of

divisible or separate performance.

2.5.5 Carroll’s three-stratum theory

This theory is hierarchical and posits a factor referred to as general intelligence, general

mental ability (GMA) or “g”, as the most abstract and general cognitive ability. Beneath “g”

reside the second stratum factors of fluid intelligence, crystallised intelligence, general

memory and learning, broad visual perception, broad auditory perception, broad retrieval

ability, broad cognitive speediness and processing speed. This theory is similar to Cattell‟s

theory of fluid and crystallised intelligence except that it posits a “g” factor at the third stratum

(Carroll, 1993).

The discussions on the various theories of intelligence indicate that there is general

intelligence “g” as a base, with other intellectual abilities that are influenced by and/or

dependent on “g”.

39

According to Cronbach (1990), some discussions in the literature seem to suggest that fluid

ability is content-less and that crystallised abilities are elementary and nonintellectual.

However, fluid ability uses knowledge in formulating and checking hypotheses. For example,

an architect designing a building, even at his or her most creative, constantly relies on a

bank of knowledge. At the crystallised end, basic language and number skills are important

(Cronbach, 1990).

With further research on the concepts of intelligence, other types of intelligence, such as

emotional intelligence, have been identified. This view is supported by other authors (e.g.

Lazear), who indicated that researchers are now looking at intelligence as a set of

capabilities that are continually expanding and changing throughout one‟s life.

Furthermore, Van Eeden and De Beer (2005, p. 121), emphasised that, “in general, we can

distinguish different types of intelligence, for instance, biological intelligence, psychometric

intelligence, and social (or emotional) intelligence”.

These views indicate the difficulties in defining the concept “intelligence”. In the current

research, however, it will suffice to accept that the concept is used in its broadest sense,

based on Cattell‟s model of fluid and crystallised intelligence.

Two issues seem to dominate current pilot selection research. Firstly, the fact that

Spearman‟s general cognitive ability (g) plays such a central role in predicting pilot success

has raised the question of whether it really makes sense to also assess specific intelligence

(Carretta & Ree, 1989; De Kock & Schlechter, 2009; Ree & Carretta, 1996; 2002). The

second issue relates to the so-called “criterion problem” (De Kock & Schlechter, 2009). Most

validity studies cite the lack of meaningful quality criteria to validate predictors against a

weakness (Hunter & Burke, 1994; Hunter & Schmidt, 1990; De Kock & Schlechter, 2009).

Regarding the first issue, according to De Kock and Schlechter (2009), proponents of this

argument argued that most specific intelligences are so saturated with g that it rarely adds

any incremental validity to batteries already containing measures of g. The authors further

stated that this argument would make sense, assuming Carroll‟s (1993) model of cognitive

ability of a hierarchy of factors with g at its apex and group factors at successive lower levels

40

to be true. For the purpose of the current research, fluid and crystallised intelligence were

assumed to be theoretically separate and were therefore discussed as such.

The next section will explore the definitions of intelligence, which will clarify the concept of

intelligence in relation to the current research.

2.5.6 Definition of intelligence

Many researchers have investigated the relationship between psychological testing and

intelligence, and it would appear that the term “intelligence” has different meanings for

different researchers.

In this section, the focus will be on understanding the term and how it is developed, taking

the theories of intelligence into account. The concept of aptitude will, for the sake of clarity,

be discussed separately. However, the stance adopted in this study, is that intelligence and

aptitude are different sides of the same coin.

Reviews of the literature on the concept of intelligence suggest that the word “intelligence” is

situation specific. According to Pyle (1979), writers such as Lierner (1977) argued that

intelligence is not the same as other psychological terms such as “learning; thinking;

problem-solving attainment or achievement”, whilst others, such as Humphreys (1971),

McFarland; 1971, stated that these terms are not qualitatively different and largely overlap.

Although intelligence as a concept has never been adequately defined, it is still regarded by

the public at large as an important entity. Confusion is widespread concerning its definition

and many still confuse IQ with intelligence (Samunda, Feuerstein, Kaufman Lewis &

Sternberg, 1998).

Moerdyk (2009) indicated that, according to the About Intelligence Newsletter, intelligence is

often seen, in the popular sense, as the general mental ability to learn and apply knowledge

to manipulate one‟s environment and to reason and have abstract thought. He further stated,

that other definitions of intelligence include adaptability to a new environment or changes in

current environment, the ability to evaluate and judge, the ability to comprehend complex

ideas and the capacity for solving problems in an original and productive way. Intelligence is

41

also regarded as the ability to learn quickly and learn from experience, and even the ability to

comprehend relationships.

Murphy and Davidshofer (2005) stated that intelligence is first and foremost a construct and

that it is impossible to formulate a definition of the actual essence of intelligence – it can be

defined only in terms of the behaviours that indicate various levels of intelligence.

According to Aiken (1991), intelligence has been viewed by educators as the ability to learn,

by biologists as the ability to adapt to the environment, by psychologists as the ability to

deduce relationships between objects and events and by information theorists as the ability

to process information.

Mundy-Castle (1974) established that the definition of intelligence in Africa was based on the

integration of both a social and a technological dimension, whereby the latter was integrated

into the former and viewed as being subordinate to the social dimension. For instance, skills

and knowledge inculcated through schooling would only be considered intelligent behaviour

insofar as such acquired cognitive abilities could later be practically implemented in the

existing social system operating within the particular culture – hence a social system which

defined intelligent behaviour in terms of rendering social services to a group as opposed to

being viewed as individualistic advancement (Segall, Dasen, Berry & Poortinga, 1999).

Armour-Thomas and Gopaul-McNicol (1998) concurred with the view of Mundy-Castle

(1974) and Segal et al. (1999), namely that the construct (intelligence) itself is a culture

invention created for the purpose of appraising who has how much of it, according to the

value system of any given society at any point in time.

According to Anastasi (1990), intelligence is not an entity in the organism but a quality of

behaviour. Intelligent behaviour is essentially adaptive, insofar as it represents effective

ways of meeting the demands of a changing environment. Anastasi (1990) further stated that

in the human species, intelligence comprises that combination of cognitive skills and

knowledge demanded, fostered and rewarded by the particular culture within which the

individual becomes socialised.

42

From the above discussion, it is clear that intelligence is an abstract with many connotations

or meanings. According to Vernon (1969), three such connotations or uses of the term are

as follows:

the innate capacity of individuals, that is, their genetic make-up

the behaviour involved in learning, thinking and problem solving

the scores obtained for intelligence tests that measure various abilities (verbal,

nonverbal, mechanical, etc.)

However, one should bear in mind is that more than a single issue is involved – intelligent

behaviour consists of a range of somewhat interrelated skills, which is what authors

emphasise when they refer to the multifaceted nature of intelligence (Pyle, 1979).

Carretta and Ree (1996) reported on a study by Olea and Ree comparing the validity of

general cognitive ability, g, and specific abilities and specific knowledge, s1 ......sn for

predicting pilot criteria, where regression equations were compared to evaluate the

predictive efficiency of g and s for each of the criteria. The above authors found that the

measure of g across all pilot criteria was the best predictor, while s contributed little. They

also reported that on the work sample criteria, the results indicated that incremental validity

of specific measures was the result of specific knowledge of aviation (e.g. aviation principles,

controls and instruments) rather than specific cognitive abilities, namely verbal, quantitative,

spatial or perceptual speed.

The implications of the divisible or separate performance of intelligence seem to suggest that

consideration should be given to how intelligence can be measured using psychological

tests. Regarding pilot selection, Ree and Carretta (1996) cautioned that generally,

intelligence should be estimated from multiple tests. Hence no estimate of intelligence

should be based on a single test.

The discussion below focuses on the measurement of intelligence, without emphasising

whether or not it is unitary based on the view that general cognitive ability (intelligence) is a

better predictor than specific abilities in pilot selection (Carretta & Ree, 1989). This should

indicate the measuring instruments that are being appropriate in the current research.

43

2.5.7 Measurement of intelligence

In conceptualising psychological testing, it is vital to introduce the aim of measurement,

which is concerned with the application of an instrument or instruments to collect data for a

specific purpose.

This is the basis of psychometric theory, the subdiscipline of industrial psychology, within

which this research is undertaken, as well as the applicable quantitative statistical

methodology. Implied in psychological testing are measurements and obtaining a record of

sample behaviour or performance which is systematic and objective enough for different

observers to make reasonably comparable findings.

According to Van Eeden and De Beer (2005), the use of intelligence measures to obtain a

score that reflects a person‟s intellectual ability has traditionally been associated with the

measurement of intelligence.

Aiken (1997) stated that, according to the tradition in psychological assessment, anything

that exists at all exists in some measurable amount. The crux of measurement, not only in

the behavioural sciences, but in all scientific disciplines, is variables. Variables, unlike

constants, are static or dynamic properties of objects or events and they have more than one

value. According to Aiken (1997), certain variables are disparate, that is, their values are

discontinuous or detached from each other, for example, male or female, while others are

continuous (i.e. in theory at least) and measurable to any desired level of accuracy.

Anastasi (1990) explained that individual differences in human intelligence can be measured

at different levels of generality or specificity, depending upon the purpose of the assessment.

She further stated that at a relatively broad level, one finds the traditional “intelligence tests”

which can be more accurately described as measures of academic intelligence or scholastic

aptitude.

These measure a kind of intelligent behaviour that is both developed by formal schooling and

required for progress along the academic ladder. For fuller assessment of an individual‟s

readiness to perform in a particular occupation or course of study, tests of separate abilities

in such areas as the verbal, mathematical, spatial, mechanical or perceptual-motor are

44

useful – and even more narrowly defined skills and knowledge may be required for the

execution of clearly defined tasks such as certain military occupational specialties (Anastasi,

1990).

Van Eeden and De Beer (2005, p. 125) cautioned that “we must remember that an IQ or

intelligence score is a theoretical concept and it cannot reflect intelligence in all its diversity”.

In other words, the score obtained is merely an inference to reflect mental abilities, based on

the purpose of the assessment.

The view of divisible or separate performance of intelligence seems to suggest that

measures to assess the specific abilities are appropriate for such assessments, while for

general intelligence broader assessment instruments will have to be used.

The relative importance of g and s in the prediction of criteria has been and remains the

centre of controversy (Calfee, 1993; Jensen, 1993; McClellan, 1993; Ree & Earles, 1992;

1993; Ree & Carretta, 1996; Schmidt & Hunter, 1998; Sternberg & Wagner, 1993;). Ree and

Carretta (1996) noted that the study of abilities indicates notable difficulties in the

interpretation of results. According to Ree and Carretta (1996), ignoring any of the following

four important issues will lead to misinterpretation:

sample size

censored samples, which restrict range

the interpretation of rotated factors

reliability, which leads to erroneous conclusions, including using tests or other

predictors so unreliably that they cannot correlate with any other measures, thus

erroneously giving the appearance of uniqueness

In this research, both the broad level of intelligence and readiness to perform pilot tasks are

assessed. The intelligence tests used as measurement instruments in this research are

Raven‟s Progressive Matrices (RMP) and the Wechsler Adult Intelligence Scale (WAIS),

together with tests of separate abilities, namely the English Literacy Skills Assessment

(ELSA), the Blox Test and the Intermediate Test Battery. However, it could be argued that

these tests have a g loading.

45

Earlier in the chapter it was indicated that aptitude would be discussed separately to provide

clarity and to conceptuale aptitude as part of intelligence. This is to ensure that a holistic

approach to intelligence is adopted for the purpose of this research.

2.6 DEFINITION OF APTITUDE

It is pertinent to mention that, according to Birch and Hayward (1994), factor analytical

theories attempt to identify the structure of abilities which make up intelligence, and a

number of more recent theories seek to explore the working of the abilities themselves. The

above authors (1994) indicated that the theories of Gardner (1983) and Sternberg (1985;

1988) are attempts to understand intelligence in terms of the complex interaction of the

various cognitive abilities and other systems.

This part of the discussion provides a framework in which aptitude is conceptualised as

special abilities under the umbrella of intelligence. The concept will be defined from a

cognitive abilities perspective, based on an integration of the notion of intelligence and

psychological testing. The researcher will endeavour to demonstrate that the

assessment/testing used by the commercial airline dealt with in the current study is used to

test for the special abilities needed to succeed in the cadet pilot training programme.

In psychology, the term “aptitude” is defined in various ways, just as there are many different

definitions of it in the English language (Kline, 1998). For instance, according to Bingham

(1937), writers use the term “aptitude” with different emphasis, some stressing inherited

capacity, others, present ability, ease of acquisition, dominant interest or some other aspect

of aptitude.

Bingham (1937) defined aptitude as a condition of a set of characteristics regarded as

symptomatic of an individual‟s ability to acquire, with training, some (usually specified)

knowledge, skill or set of responses such as the ability to speak a language, produce music,

and so forth. According to this view, however, aptitude is a present condition, a pattern of

traits, deemed to be indicative of potentialities.

According to Kline (1998), aptitude usually refers to a collection of abilities which happen to

be of value in a particular culture. Of interest here is the fact that in both definitions, nothing

46

is said about whether the “conditions or set of characteristics or collections of abilities” are

acquired or inborn.

Reschly and Robinson-Zanartu (2002) in their discussion on dynamic assessment indicated

that dynamic assessment and mediated learning intervention models presume that learning

abilities or aptitudes are modifiable rather than stable.

Murphy (1989), however, explained that abilities are relatively stable individual differences

that are related to performance on some sets of tasks, problems or other goal-oriented

activities. A distinction is made between general ability and special abilities, where the latter

refer to more specific abilities that are restricted to a relatively narrow range of goal-directed

activities (e.g. music).

According to Owen (1991), factors that are relevant in this regard are a real interest in and

preference for a particular task without which it is not possible to acquire the specific abilities

required for a certain task. This indicates that aptitude is influenced by other constructs,

which are usually not measured.

According to Reschly and Robinson-Zanartu (2002), aptitude, intelligence and achievement

as psychological constructs or type of tests are not easily distinguished. The authors stated

that the traditional distinction was that achievement tests reflected the effects of past

learning, whereas aptitude and intelligence reflected the individual‟s potential for success. In

this traditional view, both aptitude and intelligence were seen as relatively enduring traits of

the individual, not easily modified by experience or special training.

Cattell (1987) suggested that, in a sense, the term “aptitude” is a misnomer if it means to

imply that what is being measured is an inborn unchangeable characteristic. Tests of special

aptitudes were viewed as measures of special innate or hereditary talents not based on

experience.

In contrast to Bingham (1937) and his followers, Guildford (in Owen & Taljaard 1989) and

Super and Crites (1965) and others had a somewhat different view of aptitude. Owen and

Taljaard (1989) cited Guildford (1959), who referred to the term as the underlying

47

dimensions of ability. Some of these dimensions appeared to be fields of knowledge (e.g.

mechanical or mathematical).

Super and Crites (1965) adopted another approach to aptitude. It is used in either of two

ways, as when it is said that a man has a great deal of aptitude for art, meaning that he has

many of the characteristics required to be successful as an artist or when one says that a

person lacks spatial aptitude, meaning that she lacks this one specialised aptitude which is

vital in a number of occupations.

In the first instance, the term refers to a combination of traits and abilities, and in the second,

to a discrete unitary characteristic, a view proposed by Cascio (1991).

Reschly and Robinson-Zanartu (2002) indicated that the aptitude models and measures

typically focus more on intra- as opposed to inter-individual differences. The former involves

variation in the pattern of cognitive processes in the individual, whereas the latter, which is

typically used with conventional measures of intelligence and achievements, examines

differences between individuals. According to these authors, the inter-individual approaches

to interpretation are useful for determining level of performance in comparison to others,

while intra-individual approaches yield information on an individual‟s pattern of performance.

However, aptitude assessment and intervention models share the common goal of improving

academic achievement.

In the current study, the inter-individual interpretation method will be followed because

differences between individuals will provide the commercial airline with information on

individuals who are likely to succeed in completing the pilot training programme against

those who may not.

In their discussion on product versus process orientation, Reschly and Robinson-Zanartu

(2002) stated that product-oriented models focus on whether or not the individual can

correctly perform certain tasks that are assumed to be reflections of underlying aptitude. In

contrast, aptitude models that are more process oriented attempt to examine underlying

cognitive processes or thinking skills that the individual uses to achieve right or wrong

answers to the task.

48

According to Reschly and Robinson-Zanartu (2002), the process product distinction relates

to both the stability-modifiability assumption and the question of transfer of training, kind of

assessment and intended outcomes. In other words, when measuring aptitude, one

measures the outcome as reflected by the score obtained as an indication that the individual

has the aptitude to perform the task, in accordance with the product-oriented model.

Alternatively, when the scores/results are interpreted, these do measure the process the

individual uses to obtain the outcome, and this applies to the process-oriented model.

These views by Reschly and Robinson-Zanartu (2002) indicated the complexities and

interactions between intelligence and aptitude as well as whether aptitude is a unitary ability

or numerous abilities. Furthermore, it highlights the issue whether aptitude is a changeable

construct.

According to the Institute of Psychology and Edumetric Research at the Human Sciences

Research Council (Owen & Taljaard, 1989), aptitude may be regarded as the potential that

enables an individual to reach a particular level of ability with a given amount of training

and/or practice. Aptitude, together with other personality characteristics such as interest,

attitude and motivation as well as training and teaching will determine the level of proficiency

and skill a person can attain.

Van Eeden and De Beer (2005) concurred in their definition of aptitude, namely that it refers

to the individual‟s ability to acquire, with training, a specific skill or to attain a specific level of

performance.

In a study by Van Eeden, De Beer and Coetzee (2001), the aim of which was to evaluate a

battery of tests to be used as part of a process of selecting first-year students

(disadvantaged students in particular) – for engineering and other science and technology

courses at the ML Sultan Technikon in Durban, a computerised adaptive test measuring

learning potential developed specifically for disadvantaged students was included. A more

commonly used test of general ability, which measures specific aptitude, the Learning

Potential Computerised Adaptive Test (LPCAT) (De Beer 1994, 1997) was also included in

the battery. This test consists of nonverbal items only in an attempt to negate the effect of

language ability/competence on test scores. This study seems to support the view that

49

aptitude may be an indicator of an applicant‟s potential to successfully complete a training

programme when training is provided.

To avoid confusion, a point of terminology needs to be clarified here. The term "aptitude test"

has been traditionally employed to refer to tests measuring relatively homogeneous and

clearly defined segments of ability, whereas the term "intelligence test" usually refers to more

heterogeneous tests yielding a single global score such as IQ (Anastasi, 1990).

Anastasi (1990) suggested replacing the traditional concepts of aptitude and achievement in

psychometrics with the concept of “developed abilities”, that is, the level of development

attained by an individual in one or more abilities.

Any test, according to Bingham (1937), tests aptitude in so far as an individual‟s score is an

indication of future potential. Prediction value is consequently the most notable feature of an

aptitude test – without it, a test cannot be an aptitude test. This indicates that when using an

aptitude test, the purpose is to determine whether a person has the necessary susceptibility

for learning or learning ability in a particular field so that with the appropriate stimuli, he/she

will succeed in that field.

Aiken (2003) stated that because of the confusion over the difference between aptitude and

achievement, it has been recommended that the two terms be replaced with the single term

“ability” – a test of ability can be a measure of achievement or of aptitude.

Multiple aptitude tests, in contrast with general aptitude tests (i.e. intelligence tests), adopt a

differential approach to measurement of aptitude. Multiple aptitude tests provide a set of

scores for different aptitudes. These scores are used to draw an intellectual profile reflecting

the individual‟s strong and weak points (Aiken, 2003).

Aiken (2003) indicated that scores on general intelligence tests are often better predictors of

success in educational and employment situations than combined scores on measures of

special abilities. This view is supported by Vernon (1960), who concluded that general

intelligence is more important than special abilities in determining occupational success.

50

According to Van Eeden and De Beer (2005), the term “aptitude” is used to describe a

specific ability. It refers to the individual‟s ability to acquire, with training, a specific skill or

attain a specific level of performance. The authors further stated that different aptitudes and

aptitude fields are also important for success in different occupations. However, the level of

skill a person may acquire also depends on his/her general intellectual ability, interest,

personality traits, attitude, motivation and training.

It can be inferred that aptitude measurements focus on the future and achievement

measurements on the past – hence Aiken‟s (2003) statement that intelligence tests are

better predictors of success, is aligned to this research. Regarding specific intelligence (Sn),

a multitude of abilities have been found to predict pilot training success, including verbal,

quantitative, spatial and mathematical ability, as well as perceptual speed and instrument

comprehension (Carretta & Ree, 1996; De Kock & Schlechter, 2009).

For the purpose of this research, the term “aptitude” is used as a broad classification,

including verbal and mathematical ability to ensure that the constructs to be investigated are

delimited. Neither personality characteristics nor training or teaching, according to the

HRSC‟s view, fall within the scope of this research.

In this research, general intelligence tests were used as measurement instruments of

general intelligence as a unitary ability. However, according to De Kock and Schlechter

(2009), the relative importance of “g” and Sn in predicting pilot training success remains a

controversial issue.

2.6.1 Measurement of aptitude

Literature on the measurement of intelligence indicates that the concept is multifaceted,

although the factor “g” plays a significant role. As such, there are numerous psychological

tests available to measure the intellectual behaviour associated with literacy and numeracy.

Regarding aptitude, Van Eeden and De Beer (2005) commented that the difference is that

the items and subtests of measures of general cognitive functioning are selected primarily to

provide a unitary measure, and not to provide an indication of differential abilities.

51

Aiken (1997) stated that researchers need to make certain that the selected instrument has

enough range for the purpose of their investigation. In this research, the ELSA test, for

instance, was found suitable as a tool to measure verbal ability and ABET levels.

According to Owen (1996), the abilities measured by aptitude measures correspond to the

intellectual abilities measured by a test of general ability. This view is similar to that of De

Kock and Schlechter (2009), who stated that many of the additional measures that are used

are saturated with “g” and do not represent unique abilities. However, some authors (e.g.

Martinussen, 1996) disagreed and demonstrated that the inclusion of specific abilities did

have an incremental validity over and above measures of “g”.

De Kock and Schlechter (2009) pointed out that according to some authors, such as Burke,

Hobson and Linsky (1997), Carretta, Perry and Ree (1996), and Ree and Carretta (2002),

“g” remains a better predictor of pilot success than specific abilities. Other authors (e.g.

Hunter & Burke, 1994; Martinussen, 1996) came to a different conclusion and reported – as

a result of their meta-analyses – that measures of general intelligence had a low mean

validities compared to more specific measures of intelligence.

According to Bergh (2009), aptitude testing is mainly directed at assessing an individual‟s

natural potential and acquired learning, which may enable him/her to develop particular

proficiencies and skills for particular intellectual tasks.

It would seem that from the aptitude measures that an individual‟s potential can be inferred

from his/her score in these measures, which could be a reflection of his/her general mental

ability. However, Bergh‟s (2009) view appears to indicate that the measure could reflect the

potential the individual has to acquire a particular skill to perform a specific task. This view

suggests that the measure may indicate the individual‟s potential in a specific task.

Intelligence test scores often correlate highly with academic success. Numerous studies

have reported moderate to high correlations between intelligence test scores and academic

achievement. The correlations are generally highest for verbal-oriented courses. For

instance, Sternberg (1982) reported on a study by Bond, that correlations between Stanford-

Binet IQ scores and success in various high school studies ranged from 0.48 (geometry) to

0.73 (reading comprehension). This is in accordance with the view of Vernon (1969) who

52

recommended that when constructing a selection procedure for the prediction of academic

performance, it is preferable to incorporate measures of general ability measuring verbal

ability and reasoning ability.

According to Bergh (2009), most aptitude measurements are based on the following general

and group abilities, as classified by Spearman (1904) and Thurstone (1938) in particular:

intelligence – the general mental ability (g factor) to understand, learn, reason and

adapt

reasoning – the ability to think logically and find rules for solutions on given

information

memory – the ability to reproduce meaningful information such as words, symbols

and numbers obtained or retained from previous learning and experience

numerical ability – the ability to reason quickly and accurately by way of addition,

subtraction, multiplication and division

spatial ability – the ability to perceive form and space, visualising forms and distance

in two and three dimensions

verbal (language) comprehension – the ability to understand and reason about

problems containing language such as words and verbal analogies

word fluency – the ability to be fluent in word and language

perceptual speed – the ability to perceive detail and differences quickly and

accurately

Of these, verbal, spatial, reasoning and numerical abilities have been found to be best

predictors of work performance (Bergh, 2009; Hunter, 1986).

Van Eeden and De Beer (2005) appear to concur with Bergh (2009) that the abilities

measured by different aptitude measures include reasoning on the basis of verbal, nonverbal

and quantitative material, language comprehension, spatial ability, perceptual speed,

memory, coordination, and so on.

Organisations need to be highly sensitive and efficient when testing applicants for positions.

The learning organisation concept suggests a way of thinking about ensuring that

53

organisations as a system can choose to create selection batteries that will potentially

identify applicants who have a better chance of succeeding in a training programme.

According to De Kock and Schlechter (2009), the debate on the role of intelligence and

aptitude in the prediction of pilot training success is still active and can be interpreted as an

attestation to its dominance in pilot selection batteries.

The discussion seems to suggest that aptitude should be measured separately. However,

the general intelligence “g” appears to play a vital role in measuring aptitude. With the

overlap between aptitude and intelligence so apparent, the implication is to consider whether

psychological tests should measure intelligence and aptitude as a unitary measure or

separate performance measures.

2.7 PSYCHOLOGICAL TESTS THAT MEASURE INTELLIGENCE AND APTITUDE

Much has been written about psychological testing and the measurement of intelligence and

aptitude. The discussion has provided the background by defining the concepts of

intelligence and aptitude, while highlighting the development of intelligence on the basis of

theories of intelligence. At this stage it would therefore be appropriate to consider the

psychological tests that provide measures for general intelligence and special abilities or

aptitude.

The use of the mental aptitude test becomes important when endeavouring to predict a

broad range of skills. According to Cunningham (1986), a single mental ability test can be

used to predict a wide range of performances and tests that were once referred to as

intelligence tests are now having their names changed. The publishers seem to be trying to

avoid the controversy and negative connotations surrounding the term “intelligence”. The

Large-Thorndike Intelligence Test is now referred to as the Cognitive Ability Test.

A recent study by Jukes and Grigorenko (2010) on assessment of cognitive abilities in multi-

ethnic countries reported that the results of their research indicated that sub-Sahara Africa‟s

historically low levels of formal education and urbanisation are providing some kind of

explanation for the continent‟s poor performance in Western IQ tests, which are designed in

and for the context of schooling in industrialised societies.

54

Most cognitive tests used in this country assess verbal ability, numerical ability, deductive

reasoning and the like. These can largely be viewed as indicators of crystallised intelligence

(following Cattell's theory) and often concern specialised skills or knowledge promoted or

required by a given culture (Taylor, 1994). This structural approach attempts to measure

performance along dimensions assumed to constitute the fundamental structure of a domain

such as cognition (Bedell, Van Eeden & Van Staden, 1999).

Van Eeden and De Beer (2005) advocated that researchers should consider the following

factors when investigating the cross-cultural equivalence of measures:

Construct equivalent refers to the fact that one needs to investigate and ensure that

the construct being measured is equivalent across different subgroups.

Method equivalence involves referring to the way in which measures are applied,

thus ensuring that the contextual and practical application of measures does not lead

to differences between subgroups. Factors such as test sophistication and language

proficiency are relevant here.

Item equivalence ensures that different subgroups do not respond to a particular item

differently because of the way in which it has been constructed and because it does

not relate to the construct being measured.

In the light of the factors discussed above, it is possible some of the instruments used during

the cadet pilot selection process may have placed some of the applicants at a disadvantage.

Method equivalence is relevant to this research study because of the diverse nature of the

applicants with regard to educational, cultural and socioeconomic backgrounds.

Jukes and Grigorenko (2010, p. 92) cited the view by Rosselli and Ardila, namely that

nonverbal tests as being “culture free” or “culture fair”, found evidence that suggests that

education level, cultural background and urban residence are all associated with

performance on nonverbal tests. The conclusion they draw is that “that there are no „culture

free‟ or „culture fair‟ tests”.

Holburn (1993) reported that psychometric assessment techniques assess specific aspects

of a person‟s functioning. A case in point is the so-called “intelligence tests” which do not

55

measure intelligence per se. They measure the ability that is deemed important in Western

cultures, and the items, which are presented in a specific language, closely approximate

those found in Western education systems. All test scores are influenced by past learning

experiences. There are no measures of pure innate ability.

The results of a study by Carroll (1993) found that performance in a foreign language course

by students whose precourse scores in a foreign language achievement test were zero could

be predicted from their scores in a test of aptitude for learning foreign languages. As one

would expect, if the training had improved achievement but had not affected aptitude, the

scores for the achievement test would have increased significantly. However, the aptitude

test scores remained essentially unchanged.

These views suggest that caution should be exercised when measuring a construct such as

aptitude because there may be underlying factors that affect the scores. Also, the measuring

instruments should measure the constructs that have been identified as important for one to

successfully complete a pilot training programme.

According to Hesketh and Robertson (1993, p. 4), to fully understand the role of both broad

and more specialised abilities, they should both be assessed at the same time: “research on

narrow abilities may have failed because the attention has not been paid to broad abilities

and, conversely … the research on broad abilities has been hampered, because the

influence of narrow abilities has not been recognized to a sufficient degree”.

Hesketh and Robertson (1993) further stated that to achieve such an understanding, a wider

empirical and theoretical lens is needed than has been evident in the selection literature to

date, including recognition of the distinction between latent and manifest variables and the

influence of other factors, such as motivation (maximum and typical performance) and the

measurement of variables.

Implied in these views is that the variables measured during the cadet pilot selection process

should be broad enough to make meaningful deductions about the applicant‟s abilities and

potential to successfully complete the cadet pilot training programme.

56

Aiken (1997) expressed the view that general intelligence tests measure a hodgepodge of

specific aptitudes or abilities as a two-edged sword. Because intelligence tests measure an

assortment of abilities – what Cronbach (1990) referred to as bandwidth – because of their

broadness, they have proven to be moderately effective in predicting various criteria. The

complementary concept of bandwidth refers to the amount or complexity of information one

tries to communicate in a given space or time.

Having noted the significant positive correlations between measures of special abilities,

Vernon (1960) concluded that general intelligence is more important than special abilities in

determining occupational success. Whether performance varies because of differences in

work motivation or differences in the special abilities required by certain occupations, or

both, is arguable.

According to Feldhusen and Jarwan (1995), the goal of an admission programme should be

to select students who are most likely to benefit from what is offered educationally, as well as

meeting the institution's criteria of success – hence the need to identify both the

educationally advantaged and disadvantaged candidates who have the best chance of

passing a course (Nunns & Ortlepp, 1994). It is also necessary to identify those who have a

reasonable chance of passing the course with the aid of academic support programmes. The

aim of this research study is to determine whether the measurement instruments used in the

selection process could identify candidates who have the potential to successfully complete

the cadet pilot training programme.

Taylor (1994) stressed the importance of identifying those individuals who have the potential

for development, even though their abilities are currently limited by previous disadvantages.

He suggested that in order to address the inequities of the past, more emphasis should be

placed on potential as opposed to skills or specific abilities, and those individuals with high

potential should be afforded the opportunity to develop specific skills through training

programmes.

According to Van Eeden and De Beer (2005), providing a learning opportunity as part of the

assessment takes into account the fact that individuals may differ considerably in terms of

their educational and socioeconomic background. These are known to influence cognitive

57

performance. Hence individuals should be allowed to indicate their potential level of

performance regardless of their background.

Shochet (1994) stated that tests of cognitive abilities that reflect the individual's current level

of functioning rely largely on his/her previous learning experience. Groups from a

disadvantaged educational background thus typically do not perform well in such tests.

These views are relevant to this research because the applicants were from diverse

backgrounds in terms of education, socioeconomic background, culture and so forth.

Furthermore, during the selection process from which the data were obtained, the learning

potential of these applicants was not measured, which would have added value to the

decision making on which candidates would be more likely to successfully complete the

cadet pilot training programme.

It is assumed that everyone being assessed has more or less the same level of experience

with regard to the different aptitudes. This is not necessarily true in a context where people‟s

backgrounds might not be the same. The composition of the norm group with regard to

language and culture should therefore be taken into account when interpreting results (Van

Eeden & De Beer, 2005).

Numerous instruments are available to measure intelligence and aptitude. Of relevance here

is whether organisations should be measuring potential or current abilities, with due

consideration of the impact of language for those individuals whose home language is not

English.

It can be inferred that in terms of the current research this could mean that the commercial

airline could justify using mathematics, science and English as minimum criteria for selection

to participate in the cadet training programme since these subjects may provide an indication

of some of the aptitudes required for a career as a pilot.

As mentioned previously, the purpose of the current research is to evaluate the predictive

validity of the current psychological tests used to select cadet pilots to successfully complete

the commercial airline‟s cadet pilot training programme.

58

2.7.1 Requirements for psychological measures

As mentioned earlier, in South Africa, according to the requirements of the Employment

Equity Act, it is the responsibility of organisations and employers to ensure that the

psychological measures used in the selection process are reliable and valid and do not

discriminate unfairly.

According to Damos (1996), there are three basic criteria for a pilot selection battery:

The fundamental purpose of such a battery is to select individuals for the job of flying

an airplane.

The scores obtained from the battery must be reliable.

The battery must be valid.

2.7.1.1 Reliability

Foxcroft, Roodt and Abrahams (2005) and Moerdyk (2009), defined the reliability of a

measure as the consistency with which it measures what it is supposed to measure.

Moerdyk (2009) referred to four types of reliability, namely test-retest reliability, parallel or

alternate form of reliability, internal consistency and inter-score reliability. Each of these will

be briefly discussed below.

a Test-retest reliability

In determining the reliability of a measure it is necessary to administer the measure twice to

the same group of test-takers. The reliability coefficient in this case is simply the correlation

between the scores obtained on the first (X) and second (Y) application of the measure

(Foxcroft et al., 2005). According to Moerdyk (2009), the closer the correlation coefficient is

to 1, the greater the stability will be over time.

b Parallel or alternate form of reliability

In this method, two equivalent forms of the same measure are administered to the same

group on two different occasions. The correlation that is obtained between the two sets of

scores represents the reliability coefficient which is known as the coefficient of equivalence

59

(Foxcroft et al., 2005). Some theorists argue, however, that the development of a parallel or

alternate form of any test is time consuming and expensive and therefore not recommended

(Foxcroft et al., 2005; Moerdyk, 2009).

c Internal consistency or split-half reliability

This type of reliability coefficient is obtained by splitting the measure into two equivalent

halves (after a single administration of the test) and computing the correlation coefficient

between these two sets of scores. This is referred to as the coefficient of internal

consistency. Splitting the measure at its midpoint into the first and the second half is the

simplest way, but this may be problematic because measures, particularly of ability, tend to

become more difficult towards the end (Foxcroft et al., 2005; Moerdyk, 2009).

d Inter-score or inter-rater reliability

This form of reliability is concerned with the extent to which two or more raters, observers or

judges agree about what has been observed. The correlation coefficient between the scores

from two or more observers is called the inter-score reliability coefficient (Foxcroft et al.,

2005; Moerdyk, 2009).

Generally information relating to the reliability of a test is included in the test manual or

technical manual of the psychometric test (Cronbach, 1990).

2.7.1.2 Validity

According to Moerdyk (2009), there are three main forms of validity, all of which are

important, although they are applied differently in various contexts and therefore require

different kinds of evidence. The three forms are as follows:

a Construct (theoretical) validity

This is concerned with whether the assessment technique produces results that are in line

with what is already known (Moerdyk, 2009).

60

b Content validity

According to Moerdyk (2009), content validity is concerned with whether the content of the

scale or measure accurately reflects the domain it is trying to assess. This form of checking

is most appropriate for achievement and knowledge assessment.

c Criterion-related (empirical) validity

This relates the scale outcomes to some external criterion. Moerdyk (2009) describes the

following two forms of criterion-related validity:

(1) Concurrent validity. This form of validity is designed to ask whether the

measure successfully distinguishes between known groups.

(2) Predictive validity. This is focused on whether the assessment procedure can

predict how groups may differ in the future (Moerdyk, 2009).

The following additional forms of validity are worth mentioning:

(1) Face validity. According to Moerdyk (2009), the basic issue with face validity is

that the assessment technique should appear (especially to the uninformed) to be

doing what it claims to be doing.

(2) Ecological validity. This is concerned with whether the results of the assessment

are meaningful and useful outside the setting in which they are obtained.

(3) Incremental validity. According to Carretta and Ree (1996), incremental validity is

the ability of one predictor to add to the predictive efficiency of another predictor.

Damos (1996) reported that military selection batteries are overwhelmingly designed to

predict pass-fail undergraduate flight training, while Carretta and Ree (1996) referred to

these as dichotomous pass/fail training criteria. As stated in chapter one that Damos (1996)

also noted that no examination of concurrent validity of postgraduate training performance

has been made for transport pilots.

According to the internationally accepted principles and guidelines (American Psychological

Association, 2003; US Department of Labor, both cited in De Kock and Schlechter, 2009), a

sound selection procedure is one that allows valid inferences to be made regarding future

61

job behaviour from an available measured score. Muller and Schepers (2003) concurred by

stating that predictive validity refers to the degree to which a current measure (the predictor)

can predict the variable of real interest (the criterion), which is not observed until sometime in

the future (Ghiselli, Campbell & Zedeck, 1981; Huysamen, 1996). In the current research,

the researcher wished to determine whether the psychological tests measuring intelligence,

aptitude, the Matric English symbol and ABET levels (predictors) are valid in predicting the

successful completion of the cadet pilot training programme (criterion).

Huysamen (1996) asserted that predictive validation is most relevant for aptitude and

interest tests, which are used to select and classify job applicants, or applicants for

specialised training courses. According to Muller and Schepers (2003), if a test possesses

predictive validity, it improves the decision-making process, which is of particular significance

in the selection process.

At this stage, in order to contextualise the discussion thus far, it would be appropriate to

provide a brief overview of the structure and content of the cadet training programme. This

will provide a point of reference for the importance of using the appropriate psychological

tests to select candidates for the training programme.

2.8 OVERVIEW OF THE STRUCTURE AND FORMAT OF THE CADET TRAINING

PROGRAMME

The main focus of the programme is to open doors to a career that was historically closed to

previously disadvantaged individuals. The basic requirements for selection include the

following:

no more than 60 flying hours (flying experience is not a prerequisite)

compulsory Mathematics with Physical Science, Computer Science or Accounting,

and fluency in English (at level 4)

Phase 1, the preparation course, takes place over a four-month period and consists of six

modules. There are written examinations in all aspects of modules 1, 2, 4, 5 and 6 and the

required pass mark are 75% for all examinations. There are no written examinations in

module 3. After each of the six modules, all candidates are assessed individually in terms of

62

their leadership and motivation as well as having achieved an overall 75% for the written

examinations.

After 18 months of flight training, the cadets‟ graduate with a South African Commercial

Pilot‟s Licence, with Instrumentation Rating and a Frozen Airline Transport Pilot‟s Licence.

2.8.1 Content of the cadet training programme

The basic content of phase 1, the preparation course, comprising six modules, is provided in

table 2.1 below.

Table 2.1 Content of each module in phase 1: the preparation course

Module 1 Module 2 Module 3 Module 4 Module 5 Module 6

Mathematics Physical Science Computer Training

Aviation- related Geography

Life skills Whole-brain learning Business awareness Cultural diversity Managing your money Airline industry overview Customer treatment

Avionics Internal combustion engine Mechanical/gas turbine/propeller theory Electronics/radio/ instruments

Aeroplane weight and balance and performance

Radio procedural training

External providers present modules 1 and 2 over a period of seven-and-a-half weeks. It is

pertinent to mention that phase 1 of the cadet training programme as outlined above in table

2.1 covers the phonology, syntax and semantic nature of the English language and at the

same time deals with the measurements of the basic dimensions of numeracy.

Table 2.2 highlights the practical training that the cadet pilots undertake over a period of 18

months. During this phase, the cadet pilots have to do a written and practical examination.

Those candidates who do not achieve the required pass mark are afforded an opportunity to

rewrite the examination for the particular module in which they were unsuccessful.

63

Table 2.2

Content of each module of phase 2: the flying course

Module 1 Module 2 Module 3 Module 4

Meteorology Flight planning Radio Navigation

Module 5 Module 6 Module 7 Module 8

Air law

and procedures

Aircraft technical Instruments Human

performance

The practical aircraft training is provided by an accredited flying school. The training was

previously undertaken in Australia, but is now provided at a local training school in South

Africa.

2.8.2 Duties and responsibilities of a pilot

According to Loukpoulos, Dismukes and Barshi (2003), airline cockpit operations are highly

scripted. Pilots are given formal written procedures that prescribe in detail how the aircraft

must be operated in each phase of flight, and who is required to do what and in what

sequence.

A pilot‟s primary task is to operate the aircraft safely and economically. To achieve this, pilots

perform a range of tasks as depicted in table 2.3 below, with many shared between the

captain and first officer. These may vary, depending on the company, but generally include

the following:

64

Table 2.3

General duties of pilot *

Ensuring that information about the route,

weather, passengers and aircraft is correct

Analysing the flight plan, including the route

and flying altitude, to ensure that it

accurately reflects the expected

circumstances of the flight

Calculating how much fuel to take, and

supervising the loading and fuelling of

aircraft

Ensuring that all safety systems are working

properly

Briefing the cabin crew before the flight, and

maintaining regular contact with them

throughout the flight

Carrying out preflight checks on the

navigation and operating systems

Communicating with air traffic control prior

to take-off and during flight and landing

Ensuring that noise regulations are

followed during take-off and landing

Understanding and interpreting data from

the instruments and controls

Making regular checks on aircraft‟s

technical performance and position, on

weather conditions and air traffic during

flight

Communicating with passengers using

the public address system

Reacting quickly and appropriately to

environmental changes and emergencies

Updating the aircraft logbook and/or

writing a report at the end of the flight,

noting any incidents or problems in the

aircraft

*Source: Adapted from Airline pilot job description and activities (2010)

This part of the discussion covered the first phase of the cadet pilot training programme,

which is theoretical, thus forming the foundation for the second phase, which is practical and

includes the actual flying of an aircraft. On successful completion of the practical training,

the cadet pilot will obtain the Commercial Pilot Licence.

The duties and responsibilities, although not detailed, describe some of the basic functions

that a pilot must perform. It should be noted that there are differences in functions between

65

the captain and the first officer. In most instances, this is based on the individual‟s number of

flying hours. Furthermore, it will be noted that the duties mentioned do not cover the “during

the flying phase”.

2.8.3 Psychological profile of a pilot

According to Bartram and Baxter (1996), in the search to find better methods of identifying

“the right stuff”, the history of pilot selection has witnessed a wide variety of assessment

methods. The authors stated that over the past 40 or 50 years, individual differences in

general ability have remained a strong predictor of training outcome, together with an

emphasis on the importance of psychomotor coordination and various aspects of personality.

This view indicates that pilot selection has not reached consensus of what “the right stuff” is

in pilot selection, particularly for passenger airlines. Damos (1996) supported this view when

she reported that no examination of concurrent validity or postundergraduate training

performance has been made for transport pilots.

In her research, Damos (1996) reported that job analysis for operational pilots is surprisingly

difficult to find. According to her, the problem associated with defining an airline pilot‟s job is

evident in the fact that literature available on this topic dates back to 1940s and 1950s.

Identifying reliable measures of job performance is more complex in some respects for

airlines than for the military – for instance, airline pilots have tasks other than flying the

aircraft (e.g. interacting with the public), which aircraft carriers believe are a vital part of the

job and should therefore be measured (Damos, 1996).

A literature review of the profile of a commercial pilot does not provide a “definitive” profile of

a pilot. Different views have been expressed with a particular reference to the differences

between military and civil/commercial pilots relating specifically to the personality attributes.

Table 4.10 in chapter 4 provides a comparison from different sources, namely Aspeling

(1990), De Montalk (2008) and Hutchison (2009). The common theme that emerges from the

information is some measure of cognitive ability and communication.

66

2.9 CHAPTER SUMMARY

This chapter explained the concept of selection and the process followed during cadet pilot

selection. The determinants of pilot training success were also highlighted as well as the

generic duties of a pilot. The theories of Spearman (1904), Thurstone (1938, 1948), Cattell

(1987), Gustaffsson (1984) and Carroll (1993) were outlined in order to explain intelligence in

relation to aptitude, literacy and numeracy (the latter two concepts will be discussed further

in chapter 3). Concerns about the definition of intelligence were also raised.

The literature review of intelligence provides an integrated approach, thus forming a basis to

empirically test the relationship and/or correlation and differences between

intelligence/aptitude and literacy and numeracy, which will be discussed in chapter 4.

67

CHAPTER 3

LITERACY AND NUMERACY

3.1 INTRODUCTION

In this chapter the concepts of literacy and numeracy are defined and discussed in order to

operationalise these concepts in the context of the research. The literature reviews in

chapters 2 and 3 are then integrated to provide the conceptual and theoretical perspective

for the empirical study to be discussed in chapter 4.

3.2 LITERACY

According to Jukes and Grigorenko (2010), in early childhood care and education

programmes in Africa, there is increasing use of other measures of cognitive development,

which shows recognition of the importance of mother tongue as a medium of instruction.

Another view by Huysamen (1996) relating to the use of untranslated tests in English to

individuals whose home language is not English, the individual’s test performance is not only

affected by his/her standing in the construct involved , but also by his/her familiarity with the

English language and the situation depicted in the test.

These views suggest that literacy has an impact on an individual’s performance during a

selection process involving psychological tests.

Holme (2004) stated that both the Roman and Greeks saw literacy as a condition of full civic

participation and hence of citizenship itself. This confirms the view that literacy should be

viewed from a society and community perspective.

In today’s world, literacy expectations are higher and a literate person would be expected to

be able to read widely with understanding of the meaning or be able to identify the barriers to

understanding (Holme, 2004).

68

3.2.1 Literacy in the South African context

The above explanations seem to emphasise the cultural perspective based on community

norms. This means one has to look at the implications of these explanations from a South

African perspective.

Because English has become the lingua franca of business and government, a greater

responsibility now lies with employers to ensure adequate proficiency in this language

among their employees (Price, 1997). A reasonable level of English proficiency is required to

avoid conflict and misunderstanding and to promote productivity. As mentioned previously,

this improvement in language skills is likely to prove beneficial to the test performance of

many African language speakers (Bedell, Van Eeden & Van Staden, 1999).

According to Mazak (2007), the saliency of English in the domain of work and politics is not

that surprising when viewed through the lens of colonialism. Mazak (2007) also stated that

the status of English and its importance for economic success are not questioned, and these

remain unquestioned in countries throughout the world that are trying to gain access to the

world markets.

Foxcroft and Aston (2006) noted that language is one of the parameters along which cultures

vary. They (2006) further pointed out that in South Africa, 11 official languages are

recognised: nine African languages, Afrikaans and English. The implication of this view is

emphasised by Hutton (1992) who referred to a statement by Lyster that learners in South

Africa, often move to learning English as a second language (ESL) before they are fluently

literate in their mother tongue. This is partly because English is seen as the language of

education, power and work. People learn best when they can build on what they already

know, but learning to read a language one does not know is like trying to learn two systems

at once.

The current levels of school dropouts, the need to repeat grades and the high failure rates in

the National Senior School Certificate examination taken in grade 12 (formerly standard 10)

at approximately 18 years of age, all indicate considerable underachievement among black

scholars (Viljoen, 1999).

69

The results of a study by Pretorius and Naude (2002) indicated that children are ill prepared

for formal education. They reveal inadequate literacy skill, poor sentence construction, poor

sense of syntax, and inadequate sound development and knowledge of the alphabet. The

cultural strengths of these disadvantaged black children include adequate gestalt formation,

enjoyment of and proficiency in visual art, responsiveness to the concrete, expressiveness of

gestures and body language, enjoyment of and the ability to learn via hand and symbolic

representations (Pretorius & Naude, 2002). Generally, this state of inadequacy is attributed

to the different education systems pre-1994.

Foxcroft and Aston (2006) reported on a study by Herbst and Huysamen (2000) which found

that environmentally disadvantaged children assessed in a language other than that spoken

at home performed at a significantly lower level than those who were assessed in their

mother tongue. Items involving verbal comprehension were found to be biased against test-

takers who spoke an African language at home, even though they had been exposed to

English on a daily basis.

These views are confirmed by Heugh (2001), who reported that recent international studies

show that South African pupils compared extremely unfavourably with pupils in other

countries in literacy and numeracy development.

The above discussion highlights the different views on what literacy is and the role of

language, particularly English usage and reading ability. The role of literacy in formal

schooling in the South African context seems to suggest that literacy should be defined

differently. It is therefore appropriate to consider the definition as well as the theories and

models of language testing, which provide the background to the development of language

skills for basic English.

3.2.2 Definition of literacy

Moss (1994) referred to a literature review on literacy conducted by Guerra (1992), who

found no fewer than 43 definitions of the concept. These ranged from defining literacy as the

simple encoding and decoding of graphic symbols (the ability to read and write) at a basic

level to the more complex acquisition of knowledge and skills, which allow one to use

70

language according to community norms. According to the latter definition, everyone is

literate as defined by their particular community.

Wagner (1991) asserted that because literacy is a cultural phenomenon adequately defined

and understood only within the culture in which it exists, it is not surprising that definitions of

literacy may never be permanently fixed. However, he provides the following definition:

literacy is a characteristic acquired by individuals in varying degrees from just above none to

an indeterminate upper level. Some individuals are more or less literate than others, but it is

not possible to speak of literate and illiterate persons as two distinct categories.

According to Hamon (2000), some commentators have argued that literacy enables societies

to reach a higher level of intellectual achievement. Hence being literate means having a

certain sort of awareness of language (e.g. that it consists of words and sounds).

Chew (1992) saw literacy as a competence in language arts, listening, speaking, reading

and writing, with language being central to all learning and indeed all creative art.

The following definition by Unesco (United Nations Educational, Scientific and Cultural

Organisation) is appealing in its simplicity and apparent neutrality: a person is literate who

can with understanding both read and write a short simple statement on his/her everyday life

(Gillette & Ryan, in Hutton, 1992). A more recent Unesco definition emphasises that literacy

needs to be defined in relation to its uses and purposes. "A person is literate when he has

acquired the essential knowledge and skills which enable him to engage in all those activities

in which literacy is required for effective functioning in his group and community and whose

attainments in reading, writing and arithmetic make it possible for him to continue to use

these skills towards his own and the community's development" (Hutton, 1992, p. 10).

Wagner (1991, p. 68) used Gray’s (1956) definition of functional literacy, namely, “a person

is functionally literate when he has acquired the knowledge and skills in reading and writing

which enable him to engage effectively in all activities in which literacy is normally assumed

in his culture or group”.

According to Harris and Hodges (1995), because of the breadth of the concepts involved in

literacy, several investigators prefer the use of the plural “literacies”.

71

One particular influential piece of research into the life of some US communities found

literacy being used for seven socially oriented purposes, by adults and children alike (Wray &

Medwell, 1993).

Purcell-Gates (2007) emphasised that literacy is always embedded in social institutions, and

as such, is only knowable as it is defined and practised by social and cultural groups. As

such, the author suggested that literacy is best considered an ideological construct as

opposed to an autonomous skill, separable from context of use. It is preferable, as

suggested by Street in Purcell-Gates (2007), to think of the plural, literacies, instead of the

singular, literacy.

The idea that literacy is somehow associated with a different more elaborate and effective

use of language indicates that to be literate is no longer only about being able to read and

write – it is about speaking and understanding the more elaborate forms of language that

literacy has allowed us to create (Holme, 2004). He further stated that literacy would then

suggest a capacity to talk and hence to think about complicated issues and abstract

problems.

According to Holme (2004), the concept of visual literacy as the ability to interpret diagrams

and pictures adds further confusion. If literacy is partly a visual sign system it is also bound

up with an even more complex use of signs, namely language. One effect is in language

standardisation. Developing a set way to use a language with an idea of what is correct and

what is not, is very much a product of literacy. In Holme’s (2004) view therefore, literacy is

inescapably a social phenomenon.

The ability to comprehend written text is only one of the components of literacy, and

increasingly more attention is being focused on the thinking skills that enable one to utilise

knowledge to make appropriate decisions.

Molosiwa (2007) reported on a study to investigate current language and literacy issues in

the African country of Botswana. Data for the study were collected through formal interviews.

When the participants in this study were asked what literacy meant to them, they indicated

that it means the ability to read with understanding and to apply the information being read to

72

the reader’s situation, not simply the technical aspects of decoding and encoding print.

Literacy was associated with making meaning out of written material. Because the

participants in the study were graduate students and teachers by profession, their

understanding of literacy had moved beyond merely decoding words from printed text.

Hutton (1992) stated that in South Africa, definitions of literacy are complicated by the fact

that knowledge of a second language, usually English, is as vital for survival and

development as the ability to read and write in an African language. The term "literacy" is

often loosely used to include basic competency in English. Officially, the term "English

literacy" is used in South Africa to mean the language and learning needs of those people

who have between one and five years of schooling. For the purpose of the current research,

a similar meaning will be used. It can be inferred that a person will be considered literate in

terms of English literacy provided he/she has a minimum of five years of schooling.

In the South African context, the implications of this definition may mean that any individual

without formal schooling is illiterate by virtue of English being essential in all modes of

communication in speaking and writing, but this does not necessarily mean that they are

illiterate.

3.2.3 Development of language skills for basic English

The development of language skills for basic English is discussed by considering current

theories and models of these concepts. Some of the basic skills are presented in the context

of the research.

3.2.3.1 Theories and models in language testing

The basic assumption of Lado’s theory of foreign language testing (Anderson, 1976) is that

language is a system of habits of communication. This implies that the basic patterns of

language as the forms of expression are overlearnt to the extent that they become

automatic. However, Lado’s theory offers no suggestions for measuring the comprehension

difficulty of reading materials in a foreign language. Lorge (cited in Anderson, 1976) pointed

out the weakness of a practice that involved constructing comprehension tests from

passages and adopting as an index of reading difficulty the mean score obtained by a

73

sample of subjects. The difficulty of a passage could be reflected in the difficulty of the

language used in framing test items, such that a difficult passage may appear easy because

the text questions are simply phrased, or it may appear difficult because the text questions

are couched in more complex language than the language of the passage. This could be a

major source of error variance.

According to Koch (in Foxcroft & Aston, 2006), psychologists often argue that it is justifiable

to administer the measure in English, irrespective of whether English is the first or second

language of the test-taker, because it is important that the test-takers can demonstrate their

ability to perform test tasks in a language that will be used in the workplace. However, Koch

was critical of this approach for two reasons: first, it assumes that the scores on the measure

are comparable across language groups; and second, it ignores the fact that language can

be a ”nuisance factor” that impacts on the test performance of English second language

speakers.

Language may be the most important mediator of test performance, especially when the

language in which the measure is administered is not the home language of the test-taker

(Foxcroft & Aston 2006). They reported that, according to Nell (1999), it has been found that

language is one of the primary influencers of intelligence test performance, and can have a

significant negative impact when a test is administered in the test-taker’s second or third

language. During the cadet selection process, because the tests that were used were all in

English and the administration was also conducted in English, this may have impacted on

the test performance of some of the applicants.

Readability or reading difficulty is defined in terms of the degree of comprehension with

which written material is read. Reading comprehension is defined as the correspondence

between the way the writer encodes the message and the way the reader decodes it

(Anderson, 1976).

According to Anderson (1976), Shannon and Weaver’s generalised communication model,

which is seen in Taylor’s definition of cloze procedure, is linear in that there is a beginning (a

source), a transmitter (channel) and a receiver (destination). This model was developed to

deal with signal transmissions relating to the telephone.

74

Anderson (1976) cites Klemmer (1962) and McCreary and Surbarsen (1965), who cautioned

that this model was never intended as a model for human communication. The limitation of

the model when applied to a human communication system is the separation of source and

destination and transmitter and receiver. In human communication, all these functions are

combined and performed by a human being.

As stated by Anderson (1976), the language communication model is useful for the following

reasons: (1) it permits treatment in operational terms of something that is essentially

abstract; (2) it permits well-established methods of analysis to be applied, provided that the

basic assumptions are not violated; (3) it provides a framework for establishing facts; and (4)

it provides hypotheses or explains phenomena.

The theories and models of language testing provide a framework of what language tests

should measure. The discussion below focuses on some of the tests described as language

tests.

3.2.3.2 Language testing

According to Downey, Suzuki and Van Moere (2010), the assessment of spoken language

skills is crucial in professions where communication breakdowns may have catastrophic

consequences. The authors further stated that international aviation professionals, including

pilots and air traffic control personnel, are routinely engaged in communication with fellow

professionals who may not share their native (or primary) language.

This view indicates the importance of language in aviation, particularly for pilots who utilise

different airspace all over the world and need to ensure that they maintain their core function

of safety, which is paramount in civil aviation.

Making effective decisions about the language skills of aviation professionals is thus a high-

stake endeavour and depends on valid assessment practices. According to Downey et al.

(2010), this suggests that measurement tools to assess language proficiency need to be

identified and used effectively to ensure that the measures obtained provide vital information

during the selection process.

75

A testing method originally used as a readability index has recently been looked at seriously

as a possible measure of second language proficiency. Taylor termed this method the “cloze

procedure”. According to Anderson (1976), Taylor defined the cloze procedure as a method

of intercepting a message from a transmitter, mutilating its language patterns by deleting

parts of the message and administering it to the receiver, who must attempt to make the

pattern whole again, potentially yielding a considerable number of complete sentences

(units). The cloze procedure is based on the Gestalt principle of closure, which states that

there is a tendency for individuals to close incomplete forms in order to constitute wholes.

The same principle is applied in numeracy.

Oller (1979) reported on a study conducted by Hinofotis (1976) on a cloze test in open-

ended form, which was administered to incoming foreign students at the Centre for English

as a Second Language (CESL) at Southern Illinois University. The Test of English as a

Foreign Language (TOEFL) and the placement examination used at the CESL were the

written criterion measures against which the cloze test was evaluated. The data indicated

that cloze testing might indeed be a viable alternative procedure for placement and

proficiency testing. This study supports previous research in language testing which

suggests that the cloze procedure is a useful evaluation tool.

The cloze total scores on the TOEFL and CESL placement tests were strongly correlated

regardless of the scoring method used to score the cloze test. Cloze correlated with the total

TOEFL at 0.71 and with the total score on CESL, at 0.79 and 0.84 respectively. The

correlations were significant beyond the 0.05 probability level, and indicated substantial

variance overlap among the tests.

The concern with cloze procedure is the frequency of word deletions and the scoring

procedures.

The common procedure involves credit being given only if the subject replaces a blank with

the exact word that was deleted. Allowing credit (full or partial) for synonyms was reported

by Taylor (1953), Ruddel (1964), Bermuth (1965), Blamenfeld and Miller (1966) and Miller

and Coleman (1967), all cited in Anderson (1976).

76

Experimental evidence of the relationship between these different scoring procedures is in

close agreement. For instance, Anderson (1976) reported that Rankin obtained correlations

of 0.86 and 0.92 between exact replacement and synonym scores on two different cloze

tests.

Studies of aspects of message in relation to readability have as a common purpose the

provision of a set of guidelines for making writing more comprehensible. The effect of short

sentences was investigated by Coleman (1962) and Gallant (1965), as reported by Anderson

(1976), and both studies found that shorter sentences increased comprehensibility.

Anderson (1976) reported on a study by Blumenfeld (1963) that the use of active instead of

passive verbs resulted in improvements in the number of words correctly replaced.

Anderson (1976), in further studies undertaken by Ruddel (1965), found that structured

redundancy with passage controlled difficulty enhanced comprehension.

Studies, which were also concerned with message presentation and comprehension, were

undertaken by Musgrave (1963), who tested the proposition that a short paragraph

preceding a cloze passage explaining the content of the passage, increased comprehension

(Anderson, 1976). The ultimate aim of the research into the message system is not only the

prediction of language difficulty but also its control (Anderson, 1976).

Studies were conducted by Taylor (1956, 1957), as reported by Oller (1979), to explore the

relationship between the cloze scores and general cognitive abilities, attitudes and other

personality variables. Significant and uniformly high correlations were reported at the tertiary

level between cloze scores to a measure of reading. Kohler (in Anderson, 1976) applied

factor analysis to a battery of cloze tests and selected cognitive variables. He concluded

that the abilities required for the successful completion of the cloze task in order of

importance were wide range vocabulary, logical reasoning, inference, addition (representing

a speed factor) and hidden patterns (a measure of flexibility of closure).

According to Van den Berg (1994), language proficiency is the most important single

moderator of test performance because it reflects familiarity with concepts and access to the

language medium through which knowledge has to be gained.

77

In a study with Junior Aptitude Tests (JAT) constructed by Owen (1991), Van Eeden and De

Beer (2005) stated that they found differences in mean test performance for white, coloured,

Indian and black students, despite the fact that the test measured the same psychological

constructs for the these groups. They suggested that factors such as language could have

been a source of bias, especially in the case of black students who were tested in English.

From the studies presented above, it may be concluded that performance in second

language tests will show different scores resulting in certain individuals being referred to as

having an “aptitude” for certain abilities.

Oller and Hinofotis (in Oller & Perkins, 1980) provided an explanation of individual

performance through two mutually exclusive hypotheses about second language ability.

The first hypothesis, referred to as the Divisible Competence Hypothesis, postulated that

language skill is separable into components relating to linguistically defined categories, for

example, phenology, syntax and lexicon, or to the traditionally recognised skills of listening,

speaking, reading and writing.

The second hypothesis, referred to as the Unitary Competence Hypothesis, is reminiscent of

Spearman’s general factor of intelligence and postulated that second language ability may

be a more unitary factor. In fact, once the common variance of a variety of language tasks is

explained, no meaningful unique variance attributable to separate components will remain.

The Divisible and Unitary Competency Models are mentioned here because they tie in with

the detailed discussion of intelligence and special abilities/aptitude in chapter 2.

According to Downey et al. (2010), numerous assessments have been developed for the

aviation context in response to the International Civil Aviation Organisation’s (ICAO’s)

initiative, many of which utilise interlocutors in oral proficiency interview (OPI) format.

However, the OPI format faces several challenges with regard to practicality, reliability and

standardisation for widespread (global) use (Downey et al. 2010).

In their research, Downey et al. (2010) reported on a survey which showed that at least 20

aviation English tests have been developed in response to ICAO’s English-proficiency

78

requirements. The authors stated that some of these tests were developed specifically for

testing aviation professionals domestically (e.g. Japan’s English Proficiency Test for Airline

Pilots or Thailand’s Thai DCA Aviation Test), while others were developed for an

international test-taker population (e.g. Eurocontrol’s English language proficiency for

Aeronautical Communication or RMIT’s English Language Test for Aviation). The authors

further explained that most of these tests consist of listening and speaking tasks, and the

speaking section typically employs a traditional oral interview, which is scored by human

raters.

It is interesting to note that, in their study, Downey et al. (2010) referred to the fact that the

test must be reliable, valid and practical as well as the need for standardised testing

conditions across locations. This is crucial for assessment to be fair to all candidates and

these are also vital components in evaluating the usefulness of any language test. In their

discussion on reliability in spoken language testing, they stated that reliability is usually

based on how consistently two different raters assess a candidate’s performance. This

indicates that in spoken language testing the inter-rater reliability approach is used.

Regarding standardisation, Downey et al. (2010) confirmed that fair and standardised testing

procedures are critical qualities of any high stake testing, including certification of aviation

professionals. The authors also noted that studies have reported inherent problems with the

interview formats, showing that interviewers modify their language from one candidate to

another when eliciting ratable spoken performance and that elicitation skills and interviewer

variability may result in totally different ratings for the same candidates. Downey et al. (2010)

therefore stated that evidence suggests that a computer-mediated speaking test – where all

candidates receive controlled prompts from a common item pool and are scored by the same

algorithms – can be fairer than face-to-face interviews in an appropriate context.

The Versant Aviation English Test (VAET) is designed to measure facility in spoken aviation

English and common English in the aviation domain. This construct represents the ability to

understand spoken English – both within the aviation radiotelephony phraseology and topics

related to aviation (such as movement, position, time, duration, weather and animals) – and

to respond appropriately in intelligible English at a fully functional pace (Downey et al.,

2010).

79

In the current study, a locally developed language test was utilised to measure the language

proficiency of the applicants, not necessarily for aviation communication, but to assess their

language skills to determine whether they had the potential to successfully complete the

cadet pilot training programme.

3.2.3.3 Language skills for basic English

The divisible and unitary competency models suggest that certain language skills are vital for

basic English literacy. These will be discussed next with particular reference to research on

learning a second language.

Bohlmann and Pretorius (2002) reported that the persistent problem of the poor academic

performance of many students at primary, secondary and tertiary level, particularly in

science and mathematics, is disturbing. The conceptual complexity and problem-solving

nature of these disciplines make extensive demands on the reasoning, interpretive and

strategic skills of students, especially when these skills are carried out in a language that is

not the student's primary language. They further stated that there are obviously many

factors, both extrinsic and intrinsic, to a learner that contribute to poor academic

performance. A number of studies, for instance, have been conducted on the role of

language in mathematics, and much is known about the ways in which poorly developed

language skills undermine students’ mathematical performance. But to what extent does

reading competence influence a student's ability to comprehend and do mathematics? As

early as 1987 it was pointed out by Dale and Cuevas (in Bohlmann & Pretorius, 2002) that

proficiency in the language in which mathematics is taught, especially reading proficiency,

was a prerequisite for mathematics achievement.

When students have difficulty reading in order to learn, it is often assumed that their

comprehension problem stems from limited language proficiency. This reflects an underlying

assumption that language proficiency and reading ability are basically the same thing, and if

this is indeed so, then improving the language proficiency of students should improve their

comprehension. Research by Hacquebord (in Bohlmann & Pretorius, 2002) showed that this

does not readily happen. Language and reading are clearly related, but they are

conceptually and cognitively uniquely specific skills that develop in distinct ways and rely on

80

specific cognitive operations. This view seems to emphasise the divisibility and unitary

competency model.

Learning a second language is a different task as it involves learning how to express familiar

meaning, either orally or in writing, using a whole new set of language forms or structure.

Fillmore et al. (1979) emphasise this view in their assertion that the cognitive problem facing

second language learners is both immense and complex. Their explanation is that, before

the structures of the new language can be learnt, the learner needs, firstly, to comprehend

them, and secondly, somehow to gain entry into the language and then to figure out how the

pieces of it fit together. Furthermore, he/she must somehow develop fluency in the language

and ferret out its structural details, so that his/her version of it matches the one spoken by

the people around him/her (Fillmore et al., 1979).

According to Pretorius and Naude, (2002), Ross and Roe (1990) postulated that children

from homes where reading and writing are priorities; develop literacy skills more readily than

children from homes where literacy is not valued. Informal reading activities should be

interactive, implying that parents should discuss the stories, ask questions about the books

and stories and respond to children’s comments. Pretorius and Naude (2002) reported that

Heald-Taylor (1987) concluded that these types of informal reading experiences develop

children's enthusiasm for reading and love of literature, thus providing an excellent

foundation for learning to read.

Pretorius and Naude (2002) further reported that Naude (1999) investigated the language

development and language enrichment of senior toddlers in an environmentally deprived

Griqua (South African) community, and concluded that language development and

enrichment are essentially a socially mediated process. They also concluded that the

inadequate linguistics example set by both the family and the community resulted in

inadequate language development and enrichment and the senior toddlers revealed their

impoverished, undifferentiated world of language in deficiencies pertaining to mastery of

language, style of language and code of language. The results indicated that the senior

toddlers' conceptualisation was deficient and that they were concrete bound, instead of

analytical and abstract, in their language usage.

81

It can be inferred from these studies that language skills for basic English literacy will be

influenced by both the community and the culture of the learner. In a multicultural society

such as that of South Africa, barriers will exist because English is not the first language for

all cultures.

Hutton (1992), when referring to the cognitive aspects of learning a second language, stated

that learners are unable to practise and develop their higher-level cognitive skills, such as

analysis, evaluation and synthesis, because they do not have the necessary language skills.

From the above it can be concluded that the basic skills for English literacy will be influenced

by a learner’s environment and community/culture. This leads to the question of what

constitutes the components of language, which is the focus of the next section.

3.2.4 Components of language

Thus far, the discussion in this chapter has indicated a link between the environmental and

biological factors that influence the learning of a second language, particularly cognitive

development. Hence the influence of literacy and numeracy during performance in

psychological tests will have to be assessed to determine variances, if any, in the scores

obtained. The focus now will be on components of language in relation to this research.

According to Harris and Hodges (1995), a problem in defining language is that language

itself must be the medium of its own description – hence a definition of language is

conditioned by the theoretical or subjective linguistic views of the definer.

Lieberman (1975) proposed an operational definition of language as a communications

system capable of transmitting new information, whereas, Bloom and Lahey (1978) stated

that the definition of language depends on the context in which one asks the question, “What

is language?” They further described language as a code whereby ideas about the world

are represented through a conventional system of arbitrary signals.

Holme (2004, p. 132) provided the following definition of a notable linguist, (Bloomfield,

1993): ”A system of symbols used in communication, language code; in a broad sense … A

system in terms of which something can be presented by one user and understood by

82

another … A system of communication”. This definition, according to Holme (2005), included

all forms of language, human and nonhuman, technical and common.

A sociolinguist Hymes (in Holme, 2004 p.99) described language as “socially constitute”.

This meant that language should be treated as part of the larger system of communication

that creates society and makes it what it is; therefore language is a form of social action

(Holme, 2004).

Painter (cited in Hasan & Martin, 1989, p.19) indicated that “taking language as a symbolic

system implies that, whether reading or speaking, listening or writing, the individual is

engaged in making meaning in some particular context”. It can thus be inferred that there

may be other factors that contribute to the understanding of what is said and how this

information is translated to make sense to the person receiving the message.

Bloom and Lahey (1978) identified three major components of language, namely context,

form and use. These components are discussed below.

3.2.4.1 Context of language

Language context is the broader more general categorisation of the topics that are encoded

in messages. Words or signs and the relationships between them represent information or

meaning in messages. There is a distinction between topics, the particular idea encoded in

a message, also, personal and contextual, and content, which is general, depersonalised

and independent of particular context.

3.2.4.2 Forms of language

The form of utterances can be described in terms of their acoustic phonetic shape, while the

form of signed communication can be described in terms of its configurational properties.

Broadly, form in language is the means for connecting sounds or signs with meaning, and

consists of an inventory of linguistic units and the system of rules for their combination. The

essence of language is the correlation between sound (syntax – morphemes) and meaning

or experience. Associated with these forms of expression are meanings shared by all

83

members of the same cultural community. The acquisition of the forms of expression used

and the association of meaning with these forms of expression are problems faced by

subjects learning the language.

3.2.4.3 Use of language

According to Clark (1996), language is used for doing things, and it embodies both individual

and social processes. This indicates that language use is really a form of joint action that is

carried out by an ensemble of people acting in coordination with one another. Clark (1996)

further stated that language is classified by scene and medium. The scene is where

language use takes place and the medium is whether the language use is spoken, gestured

or written or printed (or mixed).

Bloom and Lahey (1978) stated that there are two major aspects of the use of language.

First, it has to do with the goals or functions of language, and second, with the influence of

linguistic and nonlinguistic context that determines how individuals understand and choose

between alternative forms of language for achieving the same or different goals.

Language use consists of the socially and cognitively determined selection of behaviours

according to the goals of the speaker and the context of the situation. Language has two

interconnected merits: first, that it is social, and second, that it supplies public expressions

for “thoughts” which would otherwise remain private (De Vito, 1973).

Clark (1996) referred to personal settings or institutional settings or mediating or private

settings. Halliday (1973) conceived of the function of language in more social terms

involving interaction, regulation and personal control.

It is important to mention that cognitive ability plays a role in language usage, a view that is

shared by Lieberman (1975), who stated that cognitive ability is a necessary factor in human

language.

This section emphasises the social mores of the community as determinants of the

components of language, particularly as described by Bloom and Lahey (1978), and stresses

the fact that cognitive ability fulfils a major role in language.

84

Much has been said about the components of language – hence discussing the dimensions

of a language will provide the foundation on which to consider the functions of a language.

3.2.5 Basic dimensions of language

The discussion in the previous sections covered the skills and components of language. The

focus below is on the dimensions of language as related to literacy.

Lankshear et al. (cited in Green 1999, p.20) referred to emerging understandings of literacy

as an articulation of language and technology viewed historically so as to take cognisance of

"marks on natural surfaces, the alphabet and other symbols systems, stylus and pencil,

printing press and the digital electronic apparatus". Basically the framework is predicated on

an integrated, holistic (or critical-holistic) view of literacy as a situational social practice and

involves understanding literacy in terms of three interrelated aspects or dimensions: the

''operational", "cultural" and "critical". The framework, in essence, involves understanding

literacy in terms of these three interlocking dimensions or aspects (Green, 1999). They can

be understood as a set of intersecting circles, bringing together language, meaning and

context (Green 1999, pp.160–163; Lankshear et al., 1997, pp. 50–52), on the one hand, and

technology, practice and context, on the other (Green, 1999).

As explained by Green (1999), the operational refers to turning "it" on, and knowing what to

do to make "it" work; the “cultural” involves using "it" to do something meaningful and

effective in particular situations and circumstances (e.g. a geography lesson, the workplace,

etc.); while the “critical” entails recognising and acknowledging that all social practices and

their meaning systems are partial and selective and shaped by power relations.

The key point is to stress the interdependence between and intricacy of these dimensions. A

critical-holistic integrated view of literacy in practice and in pedagogy addresses all three

simultaneously – none has any necessary priority, in practice, over any other or either of the

others (Green, 1999).

Importantly, however, a first-order relationship exists between the “operational” and “cultural”

dimensions of literacy learning conceived in this manner. This is in accordance with, and by

85

analogy with Halliday (1973), to the insight that learning language is learning culture, and

vice versa.

This view is aligned to that expressed by Pretorius and Naude (2002), namely that the

school syllabus, which emanates from a centralised, urban, Westernised paradigm, bears

little or no relationship to these children's lives. In the South African context, the lack of

parental involvement is exacerbated because many parents work for long periods away from

the family home. Children are often cared for by grandparents or other relatives who have

received little, if any, formal education. Children therefore tend to have difficulty when they

reach school-going age in learning a foreign language such as English. This may mean that

the teachers have to spend more time teaching English, which, in most instances, is a base

for other subjects, including numeracy.

Green (1999) reported on a study that clearly revealed that the emphasis fell heavily on the

operational dimension of literacy and computing in the schools included in the study as well

as in the system’s literature and associated policies. Little explicit attention was paid to the

cultural dimension, and even less to the critical dimension.

3.2.5.1 Basic functions of language

The previous section indicated that three dimensions of language are needed to consider the

functions of language in relation to the links of the dimensions discussed above.

There are two basic functions of language, which are termed the experiential and the

interpersonal meta-functions. Painter (in Hasan & Martin, 1989) stated that in order to fulfil

these two basic functions, language needs a further set of resources by means of which any

part of an ongoing text is linked to the rest of the text as well as to the context – this is the

textual meta-function.

Obviously, the functions of language, which manifest in the language system, occur in the

grammatical structure. The grammatical structure may in fact be regarded as the means

whereby the various components of meaning, deriving from the different functions of

language, are integrated (Halliday, 1976).

86

Halliday (1973), in line with the understanding of the functional approach as adopted in the

current, identified four models of functions of language in the context of the ideational,

interpersonal and textual components of language, as depicted in figure 3.1 below. These

models are described as follows:

the instrumental model, which refers to use as a means of getting things done

the regulatory model, which refers to the use of language to regulate the behaviour of

others

the interactional model, which refers to the use of language in the interaction between

the self and others

the personal model or the individual’s awareness of the language as a form of his/her

individuality

The functional approach, as outlined in these four models, provides the basic functions of

language in a particular context and does not explicitly indicate the impact language has

during tasks such as those experienced in psychological testing. However, this research,

which is based on the interrelationship between language, numeracy and intelligence, does

not provide an acceptable model of the impact these variables will have on the psychological

tests used to predict success in a training programme.

87

Figure 3.1. The three bases of the environment as seen from the perspective of language. Source: Adapted from Halliday & Matthiessen (1999, p. 13) According to Lieberman (1975), human language achieves a high rate of speed and

overcomes the limits of memory span by the process of encoding, which occurs two levels –

in the production of speech and in the transformational syntax of human language. Chimsy

(in Lieberman, 1975) referred to transformational syntax as the “device” that restructures the

deep underlying level of language that is subjected to semantive analysis in the actual

sentence a person writes or speaks.

Language has two primary purposes, expression and communication, where communication

is the transmission of information from one person to another. Hence the focus in this regard

is on understanding or comprehension and usage, which denote meaning around one’s own

knowledge or experience (Lieberman, 1975).

It is necessary to assume that an overlap will occur between the functions of language

because of the interrelationship between the various functions of language. In focusing on

recruitment and selection using psychological tests, a wealth of information is normally

Second-order Reality (realm of meaning)

First-order reality

Ideational

Textual

Interpersonal

88

transmitted either during the briefing of testees or during the testing itself (i.e. when the

testees respond to the test items).

It is relevant to consider the role of comprehension as it relates to this research, particularly

as a means of measurement for literacy and/or numeracy.

a Comprehension

In this section, the concept of reading, which implies comprehension, will be discussed,

followed by a review of the nature of reading, writing and vocabulary.

Word recognition Comprehension Context analysis Word meaning Phonetic analysis Explicit meaning Structural analysis Implicit meaning Sight words Aesthetic appreciative meaning Dictionary skills Study skills Figure 3.2. Components of reading Source: Adapted from Cramer (1978, p. 207) Cramer (1978) stated that it is essential that any comprehension scheme accommodates the

traditional description labels of reading comprehension skills since all texts, tests and

material use them to identify reading comprehension skills.

Components of reading

89

Cramer (1978) distinguished between four components of comprehension (see figure 3.2

above). These components are briefly discussed below:

i Comprehension of explicit meaning

This is defined as the ability to understand, at a literal level, information clearly stated in the

text. This may include locating information, following explicit directions, identifying supporting

detail, recognising stated sequences, finding explicit proof and answering factual questions.

Explicit comprehension is also referred to as obtaining or knowing the facts and may be

likened to the thinking component. Guilford (cited in Cramer, 1978) refers to this as cognitive

memory, which is the understanding and retention of information in any form as a result of all

types of experiences.

ii Comprehension of implicit meaning

Comprehension of implicit meaning may be defined as the ability to gain meaning, through

reasoning, at levels beyond explicit meaning. The processes involved are all those reading

abilities that demand thinking, productive, intellectual responses to what is being read.

Guilford (cited in Cramer, 1978) identified the following three thinking skills similar to

the comprehension of implicit meaning:

(1) convergent thinking – problem solving in situations where sufficient information is

given so that reasoning is likely to lead to a specific answer or outcome

(2) divergent thinking – problem solving when several answers are possible,

stimulated in situations where a limited amount of information is available

(3) evaluative thinking – problem solving in relation to a value judgement, stimulated

in situations where the decision is whether the available information supports a

carefully defined concept; it focuses on questions of suitability, desirability and

worthiness of ideas and information

These thinking skills require mental operations and processes.

90

iii Comprehension of word meaning

This concept is defined as the ability to understand both the connotative and denotative

meaning of words and phrases. The process includes knowledge of the denotative and

connotative meanings of words, translation of figures of speech, idioms, slang, word origin

and other circumstances relating to word meaning. Depth and breadth of word meaning

depend largely on the quality of an individual’s experience.

iv Comprehension of aesthetic appreciative meaning

This component refers to the ability to derive personal enjoyment and meaning from

materials and to understand and appreciate the literacy devices associated with

interpretation of mode, tone, beauty, humour and other affective elements associated with

the written word. It requires the use of inference and thinking and the emphasis is on the

affective rather than the cognitive side of intellect.

The above discussion of comprehension highlights the fact that the background, experience

and environment of an individual are all relevant in the development of his/her language

skills.

Table 3.1 highlights the relationship between the four components of comprehension and the

commonly used names for reading skills.

91

Table 3.1 The four basic components of comprehension

EXPLICIT (literal)

IMPLICIT (reasoning)

WORD MEANING (vocabulary)

APPRECIATIVE (affective)

Main idea (stated) Inference Denotation Mode

Factual questions Prediction Connotation Tone

Sequence (stated) Prediction Contextual meaning Characterisation

Restating information

Interpretation Synonyms Beauty

Funding proof Evaluation Antonyms Humour

Recognising details Comparison Root words Values (personal)

Recalling details Drawing conclusions Prefixes Feeling

Locating information Critical reading Metaphor Opinion

Source: Adapted from Cramer (1978, p. 206) b Reading According to Underwood and Underwood (in Cashdan, 1986), models of those cognitive

processes involved in reading and spelling emphasise the available sources of information

and the related transformational processes necessary to convert a visual symbol into a

spoken utterance or an understanding of an idea. In the course of understanding print, the

reader may be said to process or transform the information from a visual to a semantive

code. For instance, when reading aloud, visual analysis will be followed by transformation

into a sound-based code.

The recognition of a word during reading is the process whereby a purely visual stimulus is

understood as the symbol for a specific meaning. Cashdan (1986) identified two component

subskills, both involving the relationship between print and meaning, as being significant in

reading and spelling. These sub skills are as follows: the analysis of visual patterns and the

conversion of the written form of a word into a phonological form. These two processes are

92

implemented in reading as word recognition via a purely graphic route or a phonological

route and they are also implemented as spelling because of reliance on a visual and on a

phonetic pattern.

Contextual analysis refers to the context provided by the sentence, which implies that the

word itself provides visual data and the sentence provides contextual data which are of use

to the individual who is familiar with the syntax of the particular language in use.

c Writing

According to Cramer (1978), historically, oral language comes first. Since written language is

based on an oral language tradition, one would expect this relationship to influence the

attainment of literacy.

Underwood and Underwood (in Cashdan, 1986) described the skill of writing as comprising

components which include an ability to hold a pen or pencil, an ability to form letters of the

alphabet and to join them into recognisable words and sentences, and so on.

In terms of skills, producing a coherent, fluent, extended piece of writing is probably the most

difficult thing there is to do in language. To second language learners, the challenges are

enormous, particularly for those who go on to a university and study in a language that is not

their own (Nunan, 1999).

The concern from a South African perspective is the issue of communities who were

educated under various education systems, particularly Bantu education, where they were

taught to read and write in their mother tongue, but had to apply second language skills

when seeking employment and during their training in the business environment.

d Vocabulary

As part of the language system, vocabulary is intimately interrelated with grammar, and it is

therefore more than a test of target language words. Nunan (1999) stated that the

“grammaticality” of vocabulary also manifests itself in word morphology - that is, the

grammatical particles that we attach to the beginning and ends of words in order to form a

93

new word. Proponents of audiolingualism have argued that foreign language learning would

be most effective if learners concentrated their efforts on mastering the basic sentence

patterns of the language. Once these patterns have been memorised, new vocabulary can

be “slotted in” (Nunan, 1999).

Nunan (1999) further stated that proponents of a comprehension-based approach to

language acquisition argue that early development of an extensive vocabulary can enable

learners to “outperform their competence”. This means that, if one has an extensive

vocabulary, it is possible to obtain meaning from spoken and written texts, even though one

does not know the grammatical structure in which the text is encoded.

According to Larsen-Freedman (in Nunan, 1999), descriptioned grammar is an integration of

syntax (form), semantics (meaning) and pragmatics (use) which enables individuals to

communicate through language. This idea was emphasised by Bloom and Lahey (1978),

who proposed that the ability to integrate the above three major components determines a

person’s knowledge of the language.

The models, components and dimensions relating to language do not provide a holistic

approach to the understanding of language, although the above discussion indicates that

literacy is broadly inclusive of language.

Learning to read involves learning to recognise the alphabet and utilising the various letters,

firstly in monosyllablic, later polysyllabic, words before graduating to reading phrases and/or

sentences. It is pertinent to mention that both spoken and written texts contain many of the

same vocabulary items and therefore draw on the same underlying linguistic systems of

English.

Discussion of the notion that language has two functions, namely experiential and

interpersonal meta-functions, infers that text provided in a certain context and grammatical

structure will give meaning. The description of the three thinking skills as identified by

Guilford, are utilised by candidates participating in psychological tests which require reading,

writing and comprehension skills (Cramer, 1978). Hence the role of literacy, particularly in

the English language, cannot be underestimated when assessing performance in

psychological tests.

94

The first part of this chapter provided the framework of the concept of literacy. The second

part will provide the framework for the numeracy concept, thus completing the

operationalisation of these concepts as they relate to this research.

3.3 NUMERACY

The concept of numeracy is defined below to explain its application in the context of this

research. The basic components of numeracy are also discussed in terms of the

development of numeracy skills and the basic dimensions thereof.

3.3.1 Definition of numeracy

Curwin, Slater and Hart (1994) described numeracy as far more than the ability to add two

numbers together and obtain the correct answer. They defined it as knowing when to use

numbers and being confident that these numbers have meaning.

Curwin et al. (1994) further stated that numeracy involves having the following

competencies:

knowing when to use numbers and when to ignore them

knowing that the numbers used are meaningful and likely to be correct

being able to select appropriate methods to make numbers more meaningful

being able to use the results produced by others

being able to interpret results and communicate this interpretation to others

Table 3.2 highlights the major differences between arithmetic and numeracy, as identified by

Penney (in Hutton, 1992).

95

Table 3.2

Differences between arithmetic and numeracy

Arithmetic

Numeracy

Context Contrived and unlikely to occur in real life

Real problems drawn from learners' daily lives

Data Problems contain only data needed for their solution. Learners never develop the skill of selecting what is relevant

Learner extrapolates the data necessary to answer his/her needs from the mass of data that is encountered in everyday life

Techniques Mechanical Requires judgement to draw inferences at an appropriate point from the available data

Level of complexity in the operation of techniques

Operative skills Skills involved in calculation, approximation, estimation, the use of calculators and other suitable aids

Castle (in Hutton, 1992) postulated that it would appear that numeracy is more than

arithmetic, but arithmetic might be part of what constitutes numeracy. According to Glen (in

Hutton, 1992), numeracy is an integrated set of mental skills plus understanding.

Hutton (1992) stated that the Crowther Report of the Central Advisory Council for Education

(in the UK) defined numeracy in 1959 as the minimum knowledge of mathematics and

science subjects which any person should possess in order to be considered educated.

According to Castle (in Hutton, 1992), this definition implies that a sophisticated level of

mathematics and scientific understanding is required for numeracy, and that the precise level

of "minimum knowledge" is determined by the numerate or educated members of society.

The definition also implies that numeracy provides access to superior culture and

information.

Similar to literacy, it would appear that there is no single definition of numeracy. Based on

this research, the approach would be an integration of the definitions provided above.

However, the view that there is a difference between arithmetic proficiency and numeracy

will not be considered further in the current research because the above definitions do not

96

distinguish between the two concepts. For the purpose of this research, numeracy will

therefore be used as an encompassing concept.

The concept of numeracy is discussed separately from literacy to ensure that it is understood

in the context of this research. As with language, it would be appropriate to consider the

development of skills for numeracy.

3.3.2 Development of skills for basic numeracy

The literature contains many theories, models or approaches regarding what constitutes

basic skills for numeracy. The difficulty is for a researcher to identify the most relevant and

appropriate theory, model or approach for his/her particular study. Hence these theories,

models or approaches will be explained as they relate to the current research, which will

provide the foundation for exploring how the skills for basic numeracy are developed and the

dimensions of this concept.

3.3.2.1 Theories, models and skills for basic numeracy

According to Piaget (in Papalia & Olds, 1995), human beings enter the stage of concrete

operations when they can think logically about the here and now. Logical operations include

arithmetic, class and set relationships, measurement and conception of hierarchical

structures. Operations may thus be described as an advanced cognitive ability based upon

certain concepts such as, mass, weight, number, length, area volume and the ability to deal

with alterations (Papalia & Olds, 1995).

Other researchers, including Vergnaud (1997), explained certain points that were not clear in

Piaget’s theory. First, other researchers in the psychology of mathematics education

complained that Piaget did not pay enough attention to the social aspects of the teaching-

learning process. Second, the issue of the relationship between the knowing process and the

psychic and social characteristics of the situations those students are faced with is relevant.

This relates to the relationship between the problems to be solved and specific

competencies and conceptions. Last but not least, the fact that mathematical knowledge

emanates from abstracting the properties and relationships of operations and not of objects

97

may be misleading. Vergnaud’s (1997) view is that mathematics is a way to conceptualise

the real world, at least during the early stages of mathematics education.

According to Van der Zanden (1993), human beings can avoid tedious, costly, trial-and-error

experimentation by initiating the behaviour of socially competent models. This process is

referred to as observational learning, social learning and modelling (cognitive learning).

In developing conceptual thinking, a key shift may be made from learning a concept in a

concrete familiar context to understanding the concept when it is removed from that

particular concrete context and as a convention that operates across many contexts.

Mathematics (this word is used interchangeably with “numeracy”) is essentially about

disembedding and being able to see the general in the specific (Lukin & Ross, 1997). This

view is shown diagrammatically in figure 3.3 below in the model that integrates numeracy

and language when introducing numeracy as a concept that has a social purpose while

building understanding by moving from the specific to the general.

While a certain amount of learning, progress can be made with “concrete” and “street”

mathematics, the power of mathematics only really becomes available when the

generalisation becomes available (Lukin & Ross, 1997). It can thus be inferred that the

ability to generalise is highlighted as a vital skill to develop.

The basic skills of developing numeracy, as identified by Southwood, Spannerberg and

Stoker (1997), are as follows:

a Comparing, matching, sorting and classifying

These are fundamental skills based on identifying similarities and differences. Classification

is an example of detecting and expressing generality in objects, numbers, pictures and

symbols.

98

b Counting

Counting skills and understanding numbers are important in the acquisition of number facts

and in calculations. However, it is pertinent to mention that it is not enough only to know the

numbers - one also to have the ability to use them.

c Calculating

It is fundamental to be encouraged to develop and use individual methods in calculations.

This may involve the decision making in “what needs to be calculated and how best to

calculate it”.

d Estimating

Skillful estimation involves making a reasonable and accurate guess based on facts and

prior knowledge as opposed to random guessing. The ability to estimate will provide a better

appreciation of the answer to be given and may make easy detection when going wrong.

e Identifying patterns

The ability to recognise patterns helps one to master basic facts and solve problems. For

instance, numbers in bracket are calculated first.

The view that literacy and numeracy should not be separated was supported by (Halliday,

1973) who noted that numeracy is being integrated into literacy programmes because

mathematics and language are seen as interrelated social systems of meaning. Figure 3.3

below depicts the interrelationship between numeracy and literacy/language in a social

context.

99

Figure 3.3 Numeracy in a social context Source: Adapted from Lukin & Ross (1997, p. 27)

As stated by Lukin and Ross (1997), there is recognition that the development of numeracy

skills cannot be separated from the development of language and literacy skills.

According to Zevenbergen (2001), for too long, school mathematics was seen to be a

discipline different or even divorced from language. After all, there is a widely held

perception that students who do not have a strong background in English can still do

mathematics, which lends support to the notion that the two disciplines are discrete to a

significant extent. This has given rise to increasing use of word problems and highly

contextualised tasks which, to a greater or lesser extent, make links between the two worlds.

These highly contextualised tasks result in higher levels of language being used and thus

greater links between literacy and mathematics (Zevenbergen, 2001).

The social theory of numeracy views it as a meaning-making system that operates in social

context (Lukin & Roos, 1997). This view emphasises context, purpose and meaning, which

were discussed in detail under literacy above.

According to Piaget (in Hughes, 1986), in acquiring the notion of number and other

mathematical concepts, learners develop independently and spontaneously. This is done in

Tasks

Numeracy concepts &

language

Spoken & written

text

100

order to help structure and interpret their experiences, and in an effort to develop the ability

to communicate mathematically; they work cooperatively towards solving problems and so

move on to the use of correct mathematical terminology and symbols (Freedman, 2000).

Numerous writers have explained the development of basic numeracy skills, and the views

expressed seem to suggest that numeracy and language/literacy are linked and should not

be separated. In other words, the basic skills for language and numeracy are similar in that

they focus on context, purpose and meaning, which fosters the ability to communicate and

solve problems using numbers.

In order to provide context, purpose and meaning in terms of numeracy, it is necessary to

consider the basic dimensions of numeracy, which will be the focus of the next section.

3.3.3 Basic dimensions of numeracy

The focus on the basic dimensions is critical to understanding the context of the current

research. It is necessary to consider the various views on describing numeracy and

explaining the basic functions of numeracy, and these will be discussed below.

The term “addition” as used by Piaget (in Hughes, 1986) is an operation in numbers as well

as in classes, quantities and characters.

Copeland (1979) described the concept of numeracy in terms of addition and subtraction,

multiplication and division, fractions and proportions, time, chance and probability and

measurements in one, two or three dimensions.

According to Curwin et al. (1994), the dimensions of numeracy refer to addition, subtraction,

multiplication and division as basic arithmetic.

These views of the term “numeracy” seem to have a common denominator in that numeracy

encompasses addition, subtraction, multiplication and division. With this understanding, it is

now possible to consider the basic functions of numeracy.

101

3.3.3.1 Basic functions of numeracy

The ability to calculate is a fundamental mathematical skill that enables people to

communicate about human activities such as counting, comparing, measuring and

designing. It can therefore be argued that the function of numeracy is to communicate

meaning by using numbers. The discussion that follows will focus on how this meaning is

communicated.

a Addition and subtraction

Addition of numbers does not mean to increase but to group, join or rename a pair of

numbers as a simple number. Subtraction entails combining two line sections, but the minus

sign means a movement backwards for one part – in other words, subtraction is a separating

action.

Piaget (in Copeland, 1979) described addition as a reversible operation. The operation of

addition comes into being when, on the one hand, the addenda are united in a whole and on

the other, this whole is regarded as invariant, irrespective of the distribution of its parts.

The commutative property of addition means that when joining sets of objects,

using an inductive procedure, the order of adding whole numbers does not

change the sum (i.e. the associative property of addition). This means that adding

any whole numbers in whatever order does not change the sum.

b Multiplication and division

Multiplication may be regarded as quick addition, while division is really the opposite of

multiplication. According to Copeland (1979), Piaget defines multiplication of numbers as an

equidistribution. Because of their inverse relationship, multiplication and division must both

be understood, if either is to be understood.

According to Curwin et al. (1994), the rule known as the “order of operation rule”, must be

followed when using the components discussed thus far, and it is summarised as follows:

102

B – bracket

E – exponentiation (power)

D – divide

M – multiply

A – add

S – subtract

This means that one should first work out the numbers in the brackets. Exponentiation

occurs when multiplying a number by itself. Division/multiplication follow, and finally, addition

and subtraction.

c Fractions

Fractions involve understanding what one number being written over another represents. It

should be realised that the necessary characteristics of fractions, are that each part must be

the same size or equal and an integral part of the whole, which can be separated or

reassembled to form the same whole.

According to Piaget (in Copeland, 1979), seven characteristics of fractions must be complied

with prior to an operational understanding of fractions:

There cannot be thought of a fraction unless there is a divisible whole.

A fraction implies a determinate number of parts.

The subdivision is exhaustive (i.e. there is no remainder).

There is a fixed relationship between the number of parts into which the whole is

divided and the number of intersections (i.e. one cut produces two parts).

The concept of an arithmetical fraction implies that all parts are equal

When the concept of subdivision is operational, fractions have a dual character. In

other words, they are part of the original whole and they are also wholes in their own

right which may be further subdivided.

The whole remains invariant. In the principle of conservation, fractions make up the

original whole.

103

d Ratio and proportion

Ratio involves the comparison of two quantities that relate to each other. Proportion refers to

two ratios that are equivalent. The ratio concept may be applied to measurements of

volume, mass, length and calculation of area, whereas the concept of proportion is used in

the solution of problems of speed and time.

e Rate

In general, the rate is the amount by which a quantity changes when another quantity

increases by one unit.

f Measurement

The main types of measurement appropriate to the current research, in terms of operational

principles, are briefly discussed below in order to achieve the required performance.

Measurement in two or three dimensions refers to the ideas of vertical and horizontal

coordinates. Two-dimensional space involves a region or area. To locate a point in two-

dimensional space without an instrument to measure angles requires a coordinate reference

system such as horizontal and vertical axes or rectangular coordinates. This field of

measurement requires an understanding of the concept of conservation, which involves two

linear measures. The calculation area is based upon the principle of subtraction and

addition plus multiplication.

The application of two- or three-dimensional measurement is used in the construction of

diagrams in order to create a visual impact to convey details.

With a better understanding of the basic functions of numeracy, as noted in chapter 2, in

terms of the cadet pilot training programme and assessment using psychological tests, plus

the criteria to be considered for the training programme, an applicant needs to be competent

in the understanding and use of numbers. Failure to meet this minimum requirement may

have an impact on the successful completion of the cadet training programme.

104

3.4 INTEGRATION OF LANGUAGE, LITERACY AND NUMERACY

Since many texts have numeracy embedded in them, it is vital to approach texts as a whole

unit of meaning and to deal with the language, literacy and numeracy skills required to

engage successfully with them.

According to Lukin and Ross (in Halliday 1985), language has evolved to satisfy human

needs and the way in which it is organised is functional with respect to these needs. It is not

arbitrary.

Figure 3.4. Model integrating the development of numeracy concepts, skills and language

Source: Adapted from Lukin & Ross (1997, p. 47)

Extend application to further context

Develop an understanding of the interrelationship

between numeracy areas

Introduce activities that encourage learners to

Independently select and use numeracy skills

Become critical analysts

Relate to test and task

Scaffold understanding by

Moving from: specific general patterns

concrete abstract

familiar less familiar

Developing an understanding of numeracy as a tool

Developing spoken, written and symbolic language

Introduce the numeracy concept by establishing that it has a

social purpose

105

Numeracy requires systematic planning if it is to be effectively integrated into a language and

literacy programme. Hence relating numeracy concepts to text may facilitate the

understanding of mathematical concepts for social purposes. Figure 3.4 above depicts the

integration of the development of numeracy concepts, skills and language in relation to text.

In order for people to engage in issues, they often need numeracy skills to interpret

information and make considered judgements. Rogers (1975) distinguished language

competence as communicative and grammatical competence.

Grammatical competence is semantically based transformational grammar. Communicative

competence is the human abilities specific to language, based on the assumption that

linguistic capacity may theoretically be separated from cognitive capacities.

It is therefore essential for people to provide a balance of experience, which is systematic

and logical, with the application of different approaches thus helping to generate new and

more sophisticated ways of tackling word problems.

Michael (in Hutton, 1992) suggested that the languages spoken in African cultures may

retard the accessibility of mathematics concepts. In addition, the transition from the mother

tongue to English as a medium of instruction in black schools complicates concept learning.

This view seems to indicate that people whose mother tongue is not English will take longer

to understand mathematical concepts and this may contribute to their slow progress in

understanding and applying the concepts to numeracy.

Mathematical language may enable people to communicate about human activities such as

counting, comparing, measuring and designing, which will promote the development of

spoken, written and symbolic mathematical language by progressing from a basic

vocabulary requirement to language which is more formal and specialised.

Thus Grabe (in Ferdman, Weber, & Ramirez, 1994), stated that second language acquisition

cannot be envisaged as separate, independent language modalities, but instead, should be

conceived of as interdependent language skills, cognitive processes and a means of

learning. This is inclusive of numeracy.

106

According to Holme (2004, p. 13),”we need literacy to work and to make informed judgments

that participating in a democratic society requires. This understanding of social importance of

literacy was not new either, even though its incorporation into a concept of functionality was”.

In Holme’s (2004) view, functional literacy recognises that people will master the skills of

literacy to different degrees. Functional illiteracy occurs when the ability to read and write is

not sufficient for an individual to engage in society, work effectively and pursue his/her

lifestyle choices. Holme (2004) also stated that functional literacy seems to have the

advantage of assessing reading and writing skills according to a community’s socioeconomic

needs.

These views raise concern in terms of the current research in that the applicants were drawn

from diverse cultures and communities and/or societies. However, with the higher

requirements in terms of academic qualifications and specific subjects indicated earlier,

basic literacy and numeracy could be expected of this sample group. The following core

assumptions, according to Holme (2004), on which functional literacy is based, apply here:

Literacy has an economic impact.

Literacy can be measured according to what it allows us to do.

A literacy shaped by the socioeconomic opportunities it affords one is a necessary

and sufficient educational goal.

Functionality may be a useful concept for assessing the literacy of the workforce in relation

to the tasks they have to perform (Holme, 2004). This suggests that one needs to have some

understanding of the functions of the pilot, as in the case of the current research, to make

meaningful inferences from measures obtained from the assessment instrument results.

Hence measuring the literacy levels of the applicants as part of the selection process, as

was done for this research, will enhance decision making in terms of which candidates are

more likely to be successful in the cadet pilot training programme.

It can be argued based on the above literature review that the current views on thinking

(cognition), language ability and numeracy are not perceived as discrete constructs.

However, language and arithmetic abilities appear to be central to the individual’s ability to

grasp and represent reality and to solve problems in a systematic manner. Therefore, it can

107

be expected that skills in these domains would be required to perform in a wide range of

higher order areas, including successful completion of the cadet pilot training programme.

3.4 CHAPTER SUMMARY

In this chapter the concepts of literacy and numeracy were discussed and defined and linked

at the meta-level. However, they were independently described here because they consist of

various dimensions. This satisfied the second aim of the literature survey, namely to define

literacy and numeracy and identify the basic dimensions of language and numeracy as well

as the development of basic skills for English literacy and numeracy. The theories and

model underpinning second language testing, using cloze procedures, as well the

interrelationship between literacy and numeracy, were highlighted.

Various concepts were explained and defined, namely intelligence and aptitude, literacy and

numeracy, with reference to appropriate theories. From an operational perspective, literacy

was reviewed, particularly English as a second language, in order to understand the impact

on performance in psychological tests and the cadet pilot training programme.

The concepts of intelligence, literacy and numeracy were integrated in an effort to indicate

alignment of the theories, thus forming a conceptual and theoretical basis for empirically

testing the correlation between these constructs and performance in the cadet training

programme to be discussed in chapter 4.

108

CHAPTER 4

THE EMPIRICAL STUDY

4.1 INTRODUCTION

This chapter provides a detailed discussion of the research design and method used for the

empirical study. The discussion will cover the steps that were followed in the sampling,

measurement of the variables and the statistical techniques employed to analyse the data.

4.2 DESCRIPTION OF THE POPULATION AND SAMPLE The population on which the empirical study was based represents the applicants for a

commercial airline’s cadet training programme nationally, who had a confirmed level of

education of Matric (grade 12) or higher. The rationale for the cut-off in educational level was

that these applicants are expected and required to have the basic English literacy and

numeracy levels, which will enable them to read manuals and policies and understand

instructions during the training courses.

The cadet pilot selection was conducted in several stages. The applicants who met the formal

criteria in terms of age, and who had at least 12 years of education in English, Mathematics

and Science as subjects, were admitted to the first step of the selection process. They were

given several paper-and-pencil tests measuring cognitive abilities, mathematical and language

skills. A similar selection process was followed by the Norwegian Air Force, where only

candidates who met the formal criteria were admitted to the first step of the selection process

(Martinussen & Torjussen, 1998).

The initial population (n = 2 253) comprised all applicants meeting the minimum requirement

for the training programme, that is, Matric or equivalent, with English, Mathematics and

Science as subjects. It is pertinent to mention at this stage that the sample comprised all

candidates who had been selected for the pilot training in three intakes: January 2001, July

2001 and January 2002. Since the inception of the cadet training programme by the

commercial airline, no validity studies have been conducted on psychological assessment

instruments, and it was therefore deemed useful to commence with data from the initial intakes

to evaluate the relevance and appropriateness of the instruments used at that time. This view

on the use of tests for the selection of aircraft pilots was supported by Damos (1996), who

109

noted that most of the studies on pilot selection involved fighter pilots. No examination of

concurrent validity or of post-undergraduate training performance has been made for transport

pilots (Damos, 1996). She (1996) further reported that determining the criteria for airline

selection batteries is difficult - no selection data for U.S. air carriers is available in the general

literature. According to Stead (in Damos, 1996), an examination of the literature on foreign

carriers revealed only one study involving prediction of operational performance. This led

Damos (1996, p. 200) to conclude that, “at this time, no general statement can be made about

the criteria for airline selection batteries”. Based on these views, this research, using the

current data, will enable the commercial airline to enhance and improve its selection process in

the future by utilising the results of these initial psychological assessments.

This approach is supported by Muller and Schepers (2003), who asserted that many

psychologists question the use of cognitive tests in selection and feel that it is necessary to

conduct an organisation-specific criterion-related validation study before using cognitive tests.

They further stated that this view is based on the fact that the size of the validity coefficients

obtained for cognitive tests differs across many studies, often by large margins.

Table 4.1 below represents the distribution of the sample based on the half yearly intakes

during January 2001, July 2001 and January 2002.

Table: 4.1 Distribution of applications by half yearly intake

Intake Frequency Percentage

January 2001 964 42.8

July 2001 983 43.6

January 2002 306 13.6

TOTAL 2 253 100.0

The data for the three intakes were pooled and the percentage distribution of race and gender

of the pooled sample is indicated in tables 4.2 and 4.3.

110

Table 4.2

Race distribution of applicants

Table 4.2 indicates that blacks comprised the largest group of applicants followed by whites,

Asians and coloureds.

TABLE 4.3 GENDER DISTRIBUTION OF APPLICANTS

Table 4.3 shows that females were in the minority by far (15.9%). This could be ascribed to

lack of information and interest, and the fact that this has always been a male-dominated

industry. Wilson (2004) emphasised the fact that the aviation industry today is still perceived to

be a male-dominated arena. According to Wilson (2004, p. 3), “in the past, recruitment of

female aviators has often been viewed as affirmative action effort (an attempt to fill a quota)”.

Race Frequency Percentage Cumulative Percentage

Black 1 347 59.8 59.8

Coloured 122 5.4 65.2

Asian 258 11.4 76.6

White 526 23.4 100.0

Total 2 253 100.0

Gender Frequency Percentage Cumulative percent

Male 1 895 84.1 84.1

Female 358 15.9 100.0

Total 2 253 100.0

111

The sample of applicants does not reflect the racial breakdown of the South African population

and is disproportionate in terms of race and gender as depicted in table 4.4 below. Table 4.4

indicates that blacks comprised 79.5% of the total South African population, followed by whites

at 9.2%, with coloureds and Indian/Asians representing 8.9% and 2.5% respectively.

In terms of gender, the population statistics as depicted in table 4.4 show a similar picture as

per the race groups, even though the race gender distribution indicates an equal distribution.

The commercial airline is attempting to address the imbalances of the past by providing career

opportunities to previously disadvantaged race groups, while also making a meaningful

contribution in terms of employment equity. This effort was also evident in the commercial

airline’s press advertisements specifying that preference would be given to previously

disadvantaged groups.

Table 4.4

Mid-year estimated for South Africa by population group and sex, 2006

Population group

Male Female Total

Number % of

population Number % of

population Number % of

population

African 18 558 500 79.6 19 104 400 79.4 37 662 900 79.5 Coloured 2 060 000 8.8 2 138 800 8.9 4 198 800 8.9 Indian/Asian 570 200 2.4 593 700 2.5 1 163 900 2.5 White 2 138 900 9.2 2 226 400 9.3 4 365 300 9.2

Total 23 327 600 100 24 063 300 100 47 390 900 100

Source: Adapted from Statistics South Africa (2006)

From the information in table 4.5 below, it is evident that females were in the minority in all the

race groups, especially the blacks and least so the whites.

112

Table 4.5 Race by gender distribution of applicants

Race

Gender Total

Male Female

Black Count 1 182 165 1 347

% within race 87.8% 12.2% 100.0%

Coloured

Count

97

25

122

% within race

79.5%

20.5%

100.0%

Asian Count 212 45 257

% within race 82.5% 17.5% 100.0%

White Count 404 123 527

% within race 76.7% 23.3% 100.0%

Total Count 1 895 358 2 253

% within race 84.1% 15.9% 100.0%

4.2.1 Biographical profile of sample in the second phase (phase 2)

All the candidates in the first phase (phase 1) who had achieved the cut-off scores in terms of

results of the psychological tests in this phase, namely English Language Skills Assessment

(Hough & Horne, 1989), Raven’s (Raven & Court, 1985) Intermediate Battery (B/77), and two

subtests,namely Reading Comprehension and Arithmetic 1 and 2 (Wilcocks, 1973) and the

Blox (Holburn, 1992), proceeded to the second phase. The second phase consisted of the

Wechsler Adult Intelligence Scale (WAIS) (Wechsler test manual, 1997), which was then

followed by a structured interview. The number of candidates was reduced from the first to the

second phase because the intention was to select those candidates who had the potential to

be successful in the training programme and the instruments were aimed at measuring

important characteristics aligned with the “ideal” profile as outlined in table 4.10 (to be

discussed in section 4.3).

113

Table 4.6 Race distribution of second phase candidates

Race Frequency Percentage Cumulative percentage

Black 19 21.8 21.8

Coloured 15 17.2 39.1

Asian 18 20.7 59.8

White 35 40.2 100.0

Total 87 100.0

After selection, which was based on the test scores achieved (see table 4.6 above), the white

cohort (23.4% of applicant pool) were now in the majority (40.2%), with the other groups were

more or less equally represented. However, the black applicant pool declined significantly from

59.8 to 21.8%.

These results highlight the debate over whether testing is useful in a multicultural environment

such as South Africa, where it is known that disadvantaged groups tend to score lower in

certain type of assessment instruments, particularly in cognitive ability tests (Kriek &

Dowdeswell, 2007, 2010; Van Eeden et al., 2001). The psychological instruments used in

phase 1 can be clustered as cognitive ability tests, and the group differences between the

racial groups based on table 4.6, therefore seem to favour whites.

According to Kriek and Dowdeswell (2010), the occurrence of group difference can potentially

lead to indirect discrimination and have an adverse impact on previously disadvantaged

groups, depending on how the assessment instruments are used.

114

Table 4.7 Gender distribution of second phase candidates

Gender Frequency Percentage Cumulative percentage

Male 64 73.6 73.6

Female 23 26.4 100.0

Total 87 100.0

Whereas in phase 1, females made up 16% of applicants, in phase 2 (see table 4.7) they

comprised 26.4%. According to Theron (2009, p. 183), “the origin of adverse impact is

generally believed to reside in the selection instruments used for personnel selection, or

differences occurring in the latent trait being assessed”. This view seems to indicate that the

applicants and black women in particular could have experienced an adverse impact. This

could be attributed to the instruments themselves or the attributes being assessed by some of

these instruments, resulting in no black women being selected for the second phase.

As an expression of the latter view, Theron (2009) cites an example by Pyburn, Ployhart and

Kravitz (2008), who reported that the ability of organisations to simultaneously identify high-

quality candidates and establish a diverse work force may be hindered by the fact that many of

the more predictive selection procedures negatively influence the pass rate of racio-ethic

minority group members (nonwhites) and women. Pyburn et al. (in Theron (2009) stated that

unfortunately, many of the most predictive knowledge, skills, abilities and other characteristics

(KSAOs) (e.g. cognitive ability) and predictor methods (e.g. assessment centres) produce

varying degrees of mean subgroup differences, with racio-ethnic minority groups usually

scoring lower than the majority. In most realistic selection situations, these subgroup

differences are large enough to reduce employment opportunities for the racio-ethnic minority

groups and women.

According to Theron (2009), there is thus a belief that selection instruments differ in terms of

the adverse impact they have on protected groups, and the instruments thus can be graded in

terms of their relative degree of adverse impact. From these beliefs, it can be inferred that the

changes reflected in tables 4.6, 4.7 and 4.8, in terms of race and gender, may be attributed to

the psychological instruments used during the two phases of the cadet selection process

115

Table 4.8 below depicts gender per race group for phase 2. Table 4.8 Race and gender of second phase candidates

Race Gender

Total

Male Female

Black

Frequency 19 0 19

% within race 100.0% .0% 100.0%

Coloured

Frequency 14 1 15

% within race 93.3% 6.7% 100.0%

Asian

Frequency 14 4 18

% within race 77.8% 22.2% 100.0%

White

Frequency 17 18 35

% within race 48.6% 51.4% 100.0%

Total Frequency 64 23 87

% 73.6% 26.4% 100.0%

In order to further examine the final sample of approved phase 2 applicants, the percentage of

successful candidates from each demographic group was calculated. The results are

presented in table 4.9 below. It should be noted that all the applicants who met the minimum

criteria were involved in the first phase of selection, which means 100% for all race groups. In

phase 2, a percentage was calculated for those who were successful against the total race

group sample.

116

Table 4.9 Proportion of each demographic group included in phase 2, as a percentage of phase 1 participants

Race and gender

Frequency % of each group

admitted to phase 2

Applicants Phase 1

Applicants

Phase 2

Black male 1 182 19 1.6

Black female 165 0 0.0

Total black 1 347 19 1.4

Coloured male 97 14 14.4

Coloured female 25 1 4.0

Total coloured 122 15 12.3

Asian male 212 14 6.6

Asian female 45 4 8.9

Total Asian 258 18 7.0

White male 404 17 4.2

White female 123 18 14.6

Total White 526 35 6.7

Total male 1 895 64 3.4

Total female 358 23 6.4

Total candidates 2 253 87 3.9

It appears that in terms of the actual numbers, equal gender representation was achieved for

whites, but not for the other race groups. In the case of blacks, no females were selected for

phase 2, while for the coloureds, only one female was selected.

117

4.3 PSYCHOLOGICAL TESTS USED AS MEASURING INSTRUMENTS

The psychological tests used to predict the successful completion of the training programme

are supported by the literature review in chapters 2 and 3 as well as being defined by practical

issues in terms of time and resources.

It is pertinent to mention that a job analysis for a pilot was not conducted to identify the profile

and/or competencies of a cadet pilot. However, in consultation with the aviation psychologist,

who was a service provider and advisor during the initiation stages of the project, the

psychological instruments discussed below are compared to other tools that were identified,

based on a comprehensive job description developed of what a pilot would do on the job, as

discussed in section 2.8.2.

4.3.1 General proposed profile of a pilot

Table 4.10 below, as stated in section 2.8.3, provides some of the proposed requirements for a

pilot. This information was obtained from various sources, including pilot training school

recruitment requirements (Hutchison, 2009), the South African Airways’ (SAA’s) (SAA, 2010)

cadet pilot training programme, the South African Air Force (SAAF) as proposed by Aspeling

(1990) and the New Zealand airline (De Montalk, 2008).

Table 4.10 Proposed profile of a pilot

SAA cadet pilot requirement

Aspeling’s proposed profile

Hutchison’s proposed profile

New Zealand’s desirable personal attributes

Requirements

- South African citizen with passion for flying

- Medically fit

- At least 1.6

metres tall

- Pass a security

Primary cognitive abilities

- Mental alertness

- Memory

- Technical

comprehension

- Technical

knowledge

- Good

educational

qualification

- Good

communication

skills

- Level-

headedness,

calmness and

- Keenness and

enthusiasm

- Positive

attitude

- Willingness to

learn

- Awareness of

the big picture

118

check

- Good communication skills

- Good hand-eye coordination

- Preferably up to 25 years of age

- Highly motivated

- Ability to work under pressure

- Multi-tasking skills

- Confident and assertiveness

- Good judgement

- Decision-making skills

- Excellent problem- solving skills

Academic requirements

- Matric certificate/Grade 12/N3 or relevant qualification (SAQA accredited)

- Pure Maths or Statistics 101:HG symbol D or SG symbol C (level 4)

- Physical

- Arithmetic ability

- Initiative

Primary affective characteristics

- Stress

management

- Adaptability

- Emotional stability

- Assertiveness

- Institutional

aggression

- Caution

Primary psycho-physiological characteristic

- Vitality

- Alertness

- Concentration

Primary conative characteristics

- Motivation

- Responsibility

Primary psycho-social characteristics

- Leadership

- Team work

- Discipline

- Interpersonal

relations

- Communication

Primary perceptual- motor characteristics

- Complex co-

ordination

- Reaction time

the ability to

think and

respond

appropriately in

difficult

situations

- The ability to do

rapid mental

calculations

- Logical

problem- solving

- Self- confidence

and clear

speaking voice

- A disciplined

outlook and

responsible

attitude

- Good

coordination

and physical

fitness

- A sound

understanding

of science

subjects such

as Physics and

Maths

- The ability to

speak English

fluently

- The ability to

motivate and

the capacity for

team work

- Communicatio

n skills

- Command

potential

- Motivation

- Level-

headedness

- Maturity

- Ability to listen

- High personal

standards

- Good

personality

- Willingness to

learn from

mistakes

- Willingness to

conform

119

Science or Computer Science: HG symbol D or SG symbol C (level 4)

- English: HG symbol D or SG symbol C (level 4)

- Dynamic

anticipation

Primary management ability

- Mature decision

making

- Strategic planning

The selection battery used in the selection process for the cadet pilots was as follows:

English Literacy Skills Assessment (ELSA) (Hough & Horne,1989)

The Blox Test (Holburn, 1992)

Raven’s Progressive Matrices (RMP) (Raven & Court,1985)

Intermediate Test Battery (B/77), subtests: (Wilcocks, 1973)

o reading comprehension

o arithmetic 1 and 2

The Wechsler Adult Intelligence Scale (WAIS) (Wechsler, 1997)

With due consideration of the contents of table 4.10 and the psychological tests listed above in

terms of this research and from Aspeling’s proposed profile under the cluster of primary

cognitive abilities, the latter were measured using the psychological tests mentioned above

(see table 4.11). It should be noted that Hutchison’s (2009) proposed profile and SAA’s

academic requirements, namely Matric certificate/Grade 12/N3 or a relevant qualification

(SAQA accredited); Maths or Statistics 101; HG symbol D or SG symbol C (level 4); Physical

Science or Computer Science: HG symbol D or SG symbol C (level 4); and English: HG

symbol D or SG symbol C (level) were used as criteria to enter the programme and the ELSA

was used to assess the English literacy level of the applicants. The applicants’ Matric symbol

for English was also considered. Arithmetic ability (Aspeling, 1990) and the ability to do rapid

mental calculations (Hutchison, 2009) the Intermediate Battery (B/77), the arithmetic 1 and 2

subtests were used to measure these abilities. Good communication, as proposed by

Hutchison (2009) and SAA (2010) which includes written and spoken communication, as wells

as technical comprehension, as proposed by Aspeling (1990), were measured by the

Intermediate Battery (B/77) subtest, namely reading comprehension.

120

The personality attributes as proposed in table 4.10 were excluded in this research. However,

these attributes were assessed during the cadet pilot selection process.

Martinussen (1996) identified the following predictors, used in pilot selection, from 50 studies

that were included in the meta-analysis. The focus was on psychological measures and not

training experience:

cognitive tests, which were used mostly in the first screening of applicants, included

various pencil-and-paper tests measuring mechanical comprehension, spatial

orientation, time sharing, visualisation, perceptual speed, instrument comprehension

and attention

intelligence tests, which were placed in a separate category because they are intended

to measure global intelligence instead of separate abilities

psychomotor/information tests, typically measuring tracking or complex coordination

and usually requiring some sort of apparatus

aviation information tests, which consisted of questions about aviation in general and

measured motivation for flying

personality tests which included specialised tests and scales developed for aviator

selection

biographical inventories, which contained background information on the applicants

the combined index, which was a combination of several tests, namely cognitive and

psychomotor, for which apparently the correlations between the subtests and the

criteria were not reported

academic results, which were either school grades or tests that measured

mathematical or language proficiency

training experience, which involved flying experience prior to selection

This information shows that the psychological instruments used for this research were aligned

to international pilot selection processes, although not all the above-mentioned predictors were

used. Cognitive and intelligence tests, as discussed in chapter 2, biographical data, academic

results and training experience were included. Training experience was used during the paper

screening of the applications. The aviation information was assessed during the interview

phase of the selection process and the results are not reported in this research.

121

Damos (1996, p. 201) noted the following: “Job analyses for operational pilots are surprisingly

difficult to find. Indeed, no discussions of the tasks required by military transport flying were

found. The problems associated with defining an airline pilot’s job are evident in the fact that

the literature available on this topic dates from 1940s and 1950s.”

Hence the test battery used during the selection of cadet pilots was guided by information from

various sources, including other airlines as well as the air force.

The tests that were included in the selection battery to measure psychological abilities required

for a cadet pilot and which psychological attributes they measure are indicated in table 4.11

below for both phases 1 and 2 of the selection process.

Table 4.11 The psychological instruments used

Test battery What it Measures

Ph

ase

On

e

ELSA English proficiency – phonics, dictation, vocabulary, reading comprehension and verbal and numerical understanding

Blox Spatial ability, the ability to visualise three-dimensional drawings and understand the nature of the arrangements of elements in visual stimulus patterns

Raven’s Progressive Matrices

Analytical ability, deductive reasoning and the ability to integrate details into a complex whole

Intermediate Test Battery (B/77): reading comprehension

English literacy – understanding grammar and vocabulary

Intermediate Test Battery (B/77): arithmetic 1 & 2

Aptitude for figures – arithmetic calculations and arithmetic reasoning in practical situations

Ph

ase

Tw

o Test battery What it measures

Wechsler Adult Intelligence Scale (WAIS)

General Intelligence (IQ)

122

The psychological tests used during the first and second phases of the cadet selection process

are discussed providing the research findings on their suitability in predicting success in the

cadet pilot training programme

These tests serve as the independent variables for the research.

Taking into account the discussion in chapters 2 and 3 relating to psychological tests and

some of the limitations of current psychological batteries, instruments to measure both literacy

and numeracy were considered on the basis that group tests need to be conducted because of

limited resources in terms of time and capital.

4.3.2 Psychological assessment instruments used in the cadet pilot selection procedure

It is necessary at this juncture to describe the instruments used in terms of the aim,

description, administration, reliability and validity as well as the research findings.

4.3.2.1 English Literacy Skills Assessment (ELSA)

The ELSA is a measuring instrument designed and developed locally for Southern African

needs (Hough & Horne, 1989). The test is described below in terms of the aim, description,

administration, validity and reliability.

a Aim

The ELSA Intermediate is a measuring instrument that quantifies the English competency

input levels and trainability levels of employees who have to communicate, cope and

function effectively in an English language environment and/or undergo training that

utilises English tests such as books, manuals, study guides, etc., with English as the

language of learning.

b Description

The test comprises the following seven subtests:

123

Phonic skills assesses whether a learner is experiencing problems with the sound

systems of the language of learning and to what extent.

Dictation determines how well the learner “hears” English and whether the

conventions of writing are part and parcel of the learner’s literacy skills. According to

Hough and Horne (1989), spelling is also taken into account.

Basic numeracy determines whether the learner is numerate. According to Hough and

Horne (1989), numeracy is an integral part of literacy.

The asserted that a person who is literate but not numerate will not be able to look up a

telephone number, a date on a calendar or read a weather report, etc., or understand

or write down a message involving numeracy.

Language and grammar of spatial relations identify learners who have a problem

with these abilities/skills.

Reading comprehension assesses narrative writing at a relatively simple level with a

readability index of ± Grade 7 for English mother tongue users.

Cloze procedure determines exposure to and familiarity with English.

Vocabulary in context involves expository writing.

c Administration

Detailed instructions for the administration of and instructions for the test are provided in

the ELSA manual (ELSA, 1989). The marking/scoring, evaluating and reporting of the

results were provided by the supplier/developer of the assessment instrument, who

provided the organisation with reports within 72 hours of the answer sheets being

delivered.

d Reliability and validity

According to Horne (1989), the predictive validity of the test is 84% and reliability is 0.67.

The ELSA literacy levels are benchmarked against South African norms as follows:

literacy – equivalent to three years of formal schooling (mother tongue implied)

functional literacy – equivalent to eight years of formal schooling

124

academic literacy – equivalent to ten years of formal schooling

e Research

After the development of the ELSA, the developers approached the Human Resources

Research Council (HRSC) to evaluate it. The first evaluation was conducted on two sets of

test results, with the first dataset comprising 74 candidates with English as their mother

tongue and the second dataset comprising the results of 614 candidates with English as

their second language. Each dataset contained two observations of each candidate,

namely the candidate’s scholastic level and the literacy skills level as determined by the

ELSA. Schoeman (1991) reported that, for the first dataset, the contingency coefficient, C,

which is a measure of agreement between the scholastic level and literacy skills level, was

C= 0.80. It was further reported that in 37 (50%) of the cases, the literacy levels agreed

with the scholastic levels. In 61 ( 82.4%) of the cases, the literacy skills level deviated by

no more than ± 1 from the scholastic levels. There was skweness in this context in the

sense that in eight (10.8%) of cases, the deviation was +1, whereas in the 16 (21.6%) of

the cases, the deviation was – 1. On average, the literacy skills test rated the candidates

half a level (0.5) below the scholastic level, with a standard deviation of 1.3.

In case of the second dataset, it is reported that contingency coefficient C = 0.65.

Regarding the discrepancies between the literacy skills levels and scholastic levels it was

found that in 33 (5.4%) of the cases, the literacy skills agreed with the scholastic levels. In

133 (21.7%) of the cases, the literacy skills deviated by not more than ± 1 from the

scholastic level. There was a marked skewness in this context in the sense that in nine

(1.5%) of the cases, the deviation was + 1, whereas in 91 (14.8%) of the cases, the

deviation was -1. On average, the literacy skills test rated the candidates 2.6 levels lower

than the scholastic levels, with a standard deviation of 1.5.

In 1993, the HRSC was approached to determine the predictive validity of theELSA, on a

sample of 684 employees, who had passed Standard 6 (Grade 7) with English as mother

tongue, in a manufacturing sector. Kriek (1993) reported that the mean raw scores of the

respondents were 72.8%, representing a higher level of English literacy skill than was

expected from the employees who had passed Standard 6 (Grade 7). Only 14.3% of the

respondents obtained a mark below the Standard 6 (Grade 7) interval.

125

In a study to determine the influence of the language proficiency of English teachers who

are not native speakers of English on language skills of their learners, KrÜgel (2006) used

the ELSA as a measuring tool. The sample for the research consisted of eight secondary

schools in the Sedibeng East District of the Gauteng Department of Education, which had

opted to use English as a medium of instruction, where Grade 12 learners (n = 102) were

taught English by teachers (n = 9) who were not native English speakers. The researcher

concluded that the results of the ELSA indicated a clear correlation between teachers and

learners. For instance, where the teachers had good scores, the learners also had good

scores. However, where the teachers had poor scores, the learners performed even

worse. KrÜgel (2006; p. 85) cautioned that “although teachers had a profound impact on

the language skills of their learners, it must be understood that apart from teachers, there

are numerous other factors/variables (e.g. environmental influence, early stimulation,

communication, auditory problems and ineffective information processing skills, etc.) that

could have had an impact or influence on the language abilities of the learners.”

4.3.2.2 The Blox Test

The aim, description, administration, reliability and validity of the Blox Test are discussed

below.

a Aim

The Blox Test measures spatial ability and is made up of a number of factors, namely

visualisation, spatial relations and spatial orientation (BLOX Test Manual, 1983). This ability to

visualise three-dimensional drawings and understand the nature of the arrangements of

elements within visual stimulus patterns is deemed to be essential for a pilot.

b Description

This is a paper-and-pencil based test using a reusable test book. It is a nonverbal test

consisting of six examples and 45 items. The stimuli are isometric drawings of different

combinations of two, three, four, five or six cubes. Each set of cubes must be compared to

126

similar arrangements of cubes viewed from other angles. The Blox Test measures the

following:

Spatial relations and orientation. This is the ability to comprehend the nature of the

arrangements within a visual stimulus pattern primarily with respect to the examinee’s

body or frame of reference (Zimmerman, 1957). The parts that comprise the figure

maintain their relationships with one another as the whole figure is moved or rotated in

space. The examinee’s task is to recognise the same visual stimulus pattern from

different angles.

Visualisation. This is the ability to mentally manipulate, that is, rotate, invert and/or

twist one or more parts of a visual stimulus pattern and recognise the changed

appearance of the object.

c Administration

In order to ensure that the respondents have a clear understanding of what is expected of

them, the test administrator works with them through the six test examples. The respondents

must analyse each stimulus set and choose the corresponding set from the five possible

options, viewing the stimulus set from different angles. The test has a time limit of 30 minutes

and consists of 45 items.

d Reliability and validity

Holburn (1992) conducted a study of trainee engineering technicians and found that the

internal consistency data indicated correlations of consistency of 0.91 and reliability of at least

0.81 Holburn (1992).

e Research

Wheeler (1993), who used a sample consisting of 93 black male apprentices and 166 white

male apprentices employed in mines in Gauteng, reported on the regression results of the

overall predictive validity of the Blox Test. In terms of predictive power, the Blox Test makes a

significant contribution to predicting overall effectiveness with 14.87% of the variance

127

explained over and above the effects of the significant third variable race (F = 23,.9; p <

0.001).

After the significant contributions of the Blox Test and the variable mine had been partialled

out, no additional variance in overall effectiveness was accounted for by either the main (F =

0,.39; p > 0.05) or interaction effect ( F = 0.10; p > 0.05) of the variable race. Wheeler (1993)

concluded that when the Blox Test was used as a predictor of overall effectiveness, there was

no support for differential prediction in respect of either intercept or slope difference between

the two race groups.

Flotman (2002), who conducted a study on the selection of SAAF pilots between 1997 and

1999, involving 92 candidates who had completed the Ground School Phase, found that the

correlations between the Blox Test and Ground School Phase results were not statistically

significant (r = 0.091; p > 0.05). He concluded that despite the poor correlation with the Ground

School Phase results, the BloxTest remains an important component of the selection battery in

the sense that it measures one of the critical abilities a potential pilot should have, namely

spatial relations and orientation.

De Kock and Schlechter (2009) reported that the Blox Test has been shown to yield

acceptable reliability estimates of scores for various South Africa cultural groups, namely for

black Xhosa males (KR20 = 0.89, coloured Afrikaans males (KR20 = 0.82), Indian males

(KR21 = 0.79) and black Zulu males (KR21 = 0.77).

4.3.2.3 Raven’s Progressive Matrices (RPM) Test

a Aim

The RPM was developed to measure eductive ability in a way that would be minimally

contaminated by variation in the knowledge possessed by those being tested. Eductive ability

is used to refer to the process of educing or squeezing new insight and information out of that

which is perceived or already known (Raven, Raven & Court, 1998).

Effective eductive behaviour requires problem identification, reconceptualisation of the whole

field (not just "the problem”) and monitoring tentative solutions for consistency with all

128

available information (Raven et al., 1998). The RPM has also been used as a dependent

measure to evaluate the effectiveness of other programmes intended to enhance "problem-

solving ability” (Raven, et al., 1998).

b Description

The test is divided into five sets of 12 problems (sets A, B, C, D and E). Each set starts with a

problem which, as far as possible, is self-evident and develops a theme in the course of which

the problem builds on the argument of what has gone before and thus becomes progressively

more difficult. This procedure affords the respondent five opportunities to familiarise

himself/herself with the field and method of thought required to solve the problems.

Administered in a standard way, the test therefore provides a built-in training programme and

indexes the ability to learn from experience or "learning potential". The cyclical format also

affords the tester an opportunity to assess the consistency of a person's intellectual activity

across five successive lines of thinking (Raven et al., 1998).

c Administration

The key requirements are firstly to make sure that the respondents understand what they are

required to do and the method of thought required to solve the problems; and secondly, to

ensure that the tests are administered as outlined in the test manual to all who are to be tested

so that the procedure adopted corresponds to that used when collecting any reference data

with which the results will be compared.

This is a pencil-and-paper test and can be administered individually or in groups. The

respondent should be allowed to work on his/her own if he/she is capable of doing so. There is

only one example to complete. One point is scored for each correct answer using a marking

key.

d Reliability and validity

Raven et al. (1998), who reported on internal consistency, found that among young people in a

1979 British standardisation process, the correlations established for eight socioeconomic

groups ranged from 0.97 to 0.99, with the low of 0.97 being a statistical artifact.

129

According to Owen (1992), a considerable amount of work has been done with the RPM

working with job applicants in the gold mining industry. An interesting finding was that,

although the testees were not test sophisticated, the internal consistency coefficient (K-R21

plus Tucker's correction) was quite high, namely 0.87.

e Research

Numerous versions of the RPM have been developed over the years, including Raven’s

Progressive Matrices (RMP), which is a nonverbal group ability test (Raven, 1960). The

Advanced Ravens Progressive Matrices (ARMP) test was specifically developed for use with

superior adults (Raven, Court & Raven, 1988).

According to Rushton, Skuy and Fridjhon (2003), regardless of the version, Raven’s test, the

Coloured Progressive Matrices, the Standard Progressive Matrices and the ARPM are the best

known, most researched and most widely used of all culture-reduced tests (Raven, 2000).

Carpenter, Just and Shell (1990) found that the ability to deduce relations and manage a large

set of problem-solving goals tends to distinguish between high- and low-scoring subjects on

the RMP. The above authors mentioned the following reasons why the RMP is appropriate in

the study of cognitive and analytical abilities:

The large number of items included in the RMP lends itself to experimental analysis of

problem solving.

Correlations between RMP scores and other measures of intellectual achievement

suggest a general underlying construct similar to Spearman’s g–factor rather than

specific aspects of cognitive functioning.

The RMP is commonly used in research that requires language processing to be

minimised.

Several studies have concluded that the RMP measures processes central to analytical

intelligence.

According to Muller and Schepers (2003), a survey of reliability studies shows a wide range of

coefficients, from the high 0.70s to the low 0.90s. They further state that early studies revealed

a fairly high correlation between the RMP and the Stanford–Binet of r = 0.60 (Keir, 1949),

130

Wechsler performance IQ of r = 0.70 (Hall, 1957) and Wechsler verbal IQ of r = 0.58 (Hall,

1957).

Zindi (1994) compared the results of black working-class Zimbabwean children with those of

white London children, using the RPM. The English group outperformed the Zimbabwean

sample, with mean IQs of 96.71 and 72.36 respectively. The researcher suggested that the

lack of Westernised test sophistication was probably a contributory factor in this lowered

performance.

In another study, Jukes and Grigorenko (2010) noted that the RPM was used in many studies

reporting IQ scores for sub-Saharan Africans. Their study indicated that almost 50% of

intergroup differences could be attributed to the effects of schooling and urban residence.

4.3.2.4 Intermediate Battery (B77)

a Aim

The National Institute of Personnel Research (NIPR), under the auspices of the Human

Sciences Research Council (HSRC), developed the Intermediate Battery. This test battery was

designed to measure mental ability in a number of areas. It is used for vocational guidance

and also for selection of staff with more than 12 years of education (Wilcocks, 1973).

The test consists of seven subtests measuring quantitative analysis ability, verbal ability and

clerical speed and accuracy, giving an overall impression of a person's abilities and aptitudes.

The battery can be used as a whole or those tests pertaining to the ability being tested can be

used in isolation.

b Description

The battery is a paper-and-pencil test in reusable booklets. Each subtest is a unit, with its own

instructions and time limit, which means that it can be applied alone. The battery is answered

on a separate answer sheet.

131

For most of the subtests, each item takes the form of a question with a five-choice answer

from which the testee must select the correct answer.

If one considers some of the duties of a pilot, for instance, the ability to assimilate and utilise

numerical values through concentration, memory and application (e.g. the calculation of fuel

consumption based on the flight plan and weather conditions), numerical and verbal

proficiency are some of the competencies assessed.

Based on these competencies, the subtests used for the selection of cadet pilots are briefly

described below:

Arithmetic problems. This is a test of arithmetic reasoning. The 30 items are in the

form of verbal statements of computational problems in two parts A and B.

Reading comprehension: This test measures the ability to understand the content of

written passages. There are four paragraphs in the test and five questions to be

answered on each paragraph.

The time limits are 45 and 20 minutes respectively for each of the above subtests.

c Administration

Groups of fewer than 20 candidates were tested by one person. However, for larger groups,

the psychometrist was assisted by a test administrator, who distributed and collected the

testing materials, answered questions, ensured that the examples at the beginning of the tests

were correctly answered and maintained order in the testing room.

The raw scores for a particular test are those questions answered correctly by a testee. The

raw scores were verified by subtracting the number of red marks from the maximum score

possible, for example, 20 was the maximum score for reading comprehension.

d Reliability and validity

The reliability of each subtest was determined using a sample of 176 South African Railways

male clerical staff with 12+ years of education (Wilcocks, 1973).

132

The reliability coefficient for the arithmetic problems and reading Comprehension subtests

were 0.795 and 0.680 respectively (Wilcocks, 1973).

Another study was conducted with a sample of 205 South African Railways male clerical staff

with nine to ten years of education. The reliability coefficient for the arithmetic problems and

reading comprehension was 0.621 and 0.737 respectively.

e Research

A review of the available literature shows that in most instances where the intermediate battery

was used as a measurement instrument, the most commonly used subtest was the mental

alertness subtest. Considering the primary measures of this subtest (i.e. mental alertness),

which are the verbal aspects of intelligence, including reasoning tasks in the form of verbal

analogies, classification of abstract concepts and figure and letter series (Intermediate Battery

Test Administration manual, 1973), it is not surprising that this subtest was utilised. The

reading comprehension and arithmetic subtests which measure arithmetic reasoning and the

ability to understand the content of a written passage were included under the mental

alertness subtest as explained above.

Nonetheless, the researchers, Nunns and Ortlepp (1994), drew a conclusion using the mental

alertness test as a predictor of academic success in Psychology 1, the researchers found that

this test correlated statistically significantly (r = 0.37; p < 0.0001) with Psychology 1, for

educationally advantaged students, but not for educationally disadvantaged students.

4.3.2.5 The Wechsler Adult Intelligence Scale (WAIS)

a Aim

Wechsler's original intelligence test, the Wechsler-Bellevue Intelligence Scale (1939), was a

milestone in the history of intelligence testing because it incorporated both verbal and

performance scales and yielded scores for those scales in addition to an overall composite

score. Furthermore, the Wechsler-Bellevue was innovative because it provided deviation IQ

133

scores that were based on standard scores computed with the same distributional

characteristics at all ages (Wechsler, 1997).

In the technical manual (1997), it is stated that Wechsler adopted an ecological approach to

the concept of intelligence and conceived of it as a multidimensional construct, one that

manifests itself in many forms. He considered intelligence not only as a global entity but also

as an aggregate of specific abilities. Wechsler explained that intelligence is global because it

characterises the individual's behaviour as a whole. It is also specific because it is composed

of elements or abilities that are qualitatively different.

Wechsler believed that intelligence should be measured by both verbal and performance

tasks, each of which measures ability in a different way and which could be aggregated to form

a general global construct (Wechsler,1997).

b Description

In the current research, ten of the 11 subtests were used. As previously mentioned, the test

has verbal tests consisting of verbal reasoning, comprehension, arithmetic reasoning, digit

symbol forward and backward and similarities subtests. The performance/practical tests

consist of the following subtests picture completion, object assembly, block design, digits

symbol 90 and picture arrangement.

c Administration

The individual subtests will be described under the following subheadings: description and

administration and scoring.

(1) Verbal tests

Comprehension

i. Description: This test measures common-sense judgement and practical

reasoning.

ii. Administration and scoring: The manual suggests that this test should be

discontinued after four failures. In scoring this test, 1 or 0 marks are awarded

according to the generalisation and quality of the response. The total score is

20.

134

Arithmetic reasoning

i. Description: This test consists of 14 items, but testing routinely begins with the

third item since the first two are ordinarily given to respondents who fail items 3

and 4. The arithmetic scores have limited added value as measures of general

ability in the population at large, but they do reflect concentration and "ideational

discipline" (Lezak, 1995).

ii. Administration and scoring: The technical test manual (1997) suggests that the

first eight problems are given orally and the last two handed to the subject on

printed cards. Question 1 should be used as an example. Normally, the examiner

should continue until three successive problems have been failed.

The arithmetic items have a time limit ranging from 15 seconds on the first four to

120 seconds on the last one. A subject can earn raw score bonus points for

particularly rapid responses on the last four items. The maximum score is 14.

Digit span

i. Description: The digit span in the Wechsler batteries is the most common for

measuring the span of immediate verbal recall. It comprises two different tests, the

digit forward and digit backward.

Both tests consist of seven pairs of random number sequences that the examiner

reads aloud at the rate of one per second. Both tests thus involve auditory

attention. In addition, both depend on a short-term retention capacity.

ii. Administration and scoring

Digit forward: The examiner requests the subject to repeat the numbers in the

same sequence in which the examiner said them. The test will normally be

discontinued after the subject has failed both series at any level, but the examiner

should proceed with the longer series if there are any indications of varying

attention.

135

Digit backward: The examiner requests the subject to repeat the numbers given

backwards. Generally, the examiner should pronounce the digits at a rate of one

per second. When a sequence is repeated correctly, the examiner reads the next

longer number sequence, continuing until the subject fails a pair of sequences or

repeats a nine-digit series correctly.

The test manual indicates that the score is the highest number of digits repeated

without error on either of the two trials, except that no credit is given for success

following on a failure on both trials. The maximum total score for the digit forward

and backward is 17.

Combining the two digit span tasks scores to obtain one score, which is the score

that is entered for statistical analysis of the Wechsler tests, indicates that the two

tests are treated as if they measured the same behaviour or are very highly

correlated behaviours (Lezak, 1995).

Similarities

i. Description: In this test of verbal concept formation, the testee must explain

what each of the pair of words has in common.

ii. Administration and scoring: The word pairs range in difficulty from the simplest

(Orange-Banana), which only retarded or impaired adults fail, to the most difficult

(Fly-Tree) in the WAIS and (Praise-Punishment) in the WAIS-R.

The test manual stipulates that responses are scored 2, 1 or 0, depending on the

degree and quality of the generalisation. The guide to marking must always be

used in evaluating responses. The maximum score is 24.

(2) Practical or performance tests

Picture completion

136

i. Description: The test consists of 15 drawings, each of which has a part missing.

Incomplete pictures of human features, familiar objects or scenes are arranged in

order of difficulty with instructions to explain which important part is missing.

ii. Administration and scoring: The examiner presents the cards in numerical order

and the subject has to name or indicate the missing part in each.

One point for each picture for which a correct response is given within the time limit.

No half marks are given. The maximum score is 15.

Object assembly

i. Description of the test: Lezak (1995) describes object assembly as a test of speed

of visual organisation and motor response because it tests the capacity for visual

organisation. Visual acuity and dexterity also make significant contributions.

This test consists of three items, namely a manikin, a profile and a hand.

ii. Administration and scoring: The test manual stipulates that the order of

presentation of the objects is a manikin, a profile and a hand.

All responses are scored for both time and accuracy. The manikin has a time limit

of two minutes, while the profile and hand have three minutes. Although the

manikin has a time limit, it is scored for accuracy only, whereas the profile and

hand are scored for both time and accuracy. This means that the final score for

these two is the sum of accuracy and time credits. The total score for the three

objects is a maximum of 26.

Block design

i. Description: This test is a construction test in which the subject is presented

with red and white blocks, four or nine, depending on the item. Each block has two

white and two red sides, and two half-red half-white sides with colours divided

along the diagonal.

ii. Administration and scoring: The test material consists of a box with 16 cubes

and nine design cards, with the first two cards being samples. The examiner

137

should remove four cubes from the box and display them to the subject and

describe the blocks.

The four-block designs have one-minute time limits and the nine-block design a

two-minute limit. The subject can earn one or two bonus points for speed on the

last designs of the WAIS. After each design has been completed, the examiner

must break the design up, leaving one block facing the right way up for the next

design, before presenting the next card.

The test is scored for both accuracy and time. Three points are given for each

design correctly reproduced within the maximum time limits. No credit for time is

allowed unless the design is reproduced just as shown on the card. Blocks need

not fit perfectly, but the design must be absolutely correct. The maximum score is

42.

Digit symbol

i. Description of the test: The symbol substitution task is printed in the WAIS test

booklet. The task is a measure of perceptual and graph motor speed. The

subject's performance on this task is a means of determining skills that might

have affected his/her score on digit symbol coding. According to Lezak (1995),

for most adults, the digit symbol is a test of psychomotor performance. Motor

persistence, sustained attention, response speed and visuomotor coordination

play a key role in a normal person's performance, but visual acuity does not.

ii. Administration and scoring: Digit symbol substitution requires the subject to

copy the symbols.

The examiner places the digit symbol sheet before the subject and indicates the

key at the top, with an explanation that the subject must complete the space

underneath with a mark that corresponds to the number at the top.

The score is the total number of symbols correctly entered. Precision and neatness

are disregarded, but recorded symbols must be identifiable as keyed symbols. The

maximum score is 67.

138

d Reliability and validity

Corrected split-half reliability coefficients for verbal IQ (.95 to .97) and full scale IQ (.96 to .98),

and their respective standard errors of about two points, are quite acceptable. The reliability of

performance IQ is excellent, averaging .93, although the value of .88 at ages 16 to 17 is not

ideal. Subtest reliability coefficients average values exceed .80 for nine of the 11 subtests.

Only picture arrangement (.74) and object assembly (.68) fall short of expectations. Test-retest

reliability coefficients affirm the excellent reliability of the verbal and full scales, and show

performance IQ to be quite acceptable (.89 to .90). Test-retest coefficients for the subtests

confirm the reliability of all tasks except object assembly and picture arrangement

e Research

Foxcroft and Aston (2003) reported on a study conducted by the task team responsible for the

adaption of the WAIS-III for English-speaking South Africans, who obtained the following mean

scores on the English proficiency test: 112.62 (SD = 9.75) for English-speaking whites, 106.52

(SD = 11.19) for English-speaking coloureds, 102.72 (SD = 1.50) for Afrikaans-speaking

whites, 99.72 (SD = 13.56) for English-speaking blacks, 95.88 (SD = 14.75) for Afrikaans-

speaking coloureds and 90.37 (SD = 17.47) blacks who spoke English at work but for whom

English was largely a second or third language.

When the mean difference were statistically compared, it was found that the mean score for

English-speaking whites was significantly higher than that of Afrikaans-speaking whites (p <

0.001), English-speaking coloureds (p = 0.0012), Afrikaans-speaking coloureds (p < 0.001),

English-speaking blacks (p < 0.001) and blacks who spoke English at work, although this was

not their first language (p < 0.001) (Foxcroft & Aston, 2003). The authors concluded that the

groups differed significantly in terms of their English proficiency levels, with groups for whom

English was a second (or third) language, scoring 0.5 to almost two standard deviations below

the white English first language sample.

In South Africa, Shuttleworth-Jordan (1996) conducted a study drawing on African first

language students studying in the medium of English at a traditionally white university. In this

study, there was evidence of consistent but only marginally lower scores for black African first

language versus white English first language students on a variety of neuropsychological

139

measures, including the South African Wechsler Adult Intelligence Scale’s (SAWAIS’s) digit

span and digit symbol subtests. Furthermore, it was found that the black African first language

scores were equivalent to those of comparable US standardised data.

According to Shuttleworth-Edwards et al. (2004), when comparing the indications from the

Avenant and Shuttleworth-Jordan studies, which both focused on black university students,

using the digit span and digit symbol subtests of the SAWAIS, it appeared that the quality of

education should be used to explain the highly discrepant finding. This is view is based on the

Avenant (1988) study in which the WAIS-R was applied to a sample comprising black prison

wardens and students from historically black universities. It was found that the university

undergraduate students scored better than the prison wardens, but fell significantly below the

US standardised sample, with a full scale IQ of 77 (recalculated in Nell, 1999, p. 132), while

the prison warden’s mean full scale IQ was 73.

Albeit ostensibly equivalent for educational level, in Avenant’s (1988) study, students from the

historically disadvantaged black universities demonstrated an IQ level in the borderline range.

In contrast, in the Shuttleworth-Jordan’s study, the students from historically advantaged white

universities achieved scores that were commensurate with the US standardisation data, and

that were not substantially different from their white English-speaking university counterparts

(Shuttleworth-Edwards et al., 2004).

4.3.2.6 Biographical factors and matriculation results

One of the requirements for the applicants for the cadet pilot programme is that they should

have a matriculation certificate or equivalent, with a pass in Maths or Science and English.

Their race and gender are also considered. These requirements, as indicated in chapter 1,

were stipulated to ensure that the commercial airline achieves its affirmative action targets as

well as diversity in its workforce.

In the discussion on literacy and numeracy, which was covered in chapter 2, the importance of

language was explored in detail. Van Rooyen’s (2001) study on home language indicated that

a person’s home language was a significant predictor of academic performance, (p < 0.01),

and it explained 10.8% of the variance in the mean of percentage mark (MCPM).

140

A literature search on biographical factors with regard to gender revealed conflicting results.

For instance, a study by Everette (1991) on 1 695 students in Australia in 1987, showed that

there is a tendency for women to have a higher pass rate than men (Everette, 1991).

Whereas, Zaaiman (1998) reported on a study by Brusselmans-Dehairs and Henry (1994),

who demonstrated that boys generally performed better than girls in mathematics.

Many organisations use Matric results as a minimum criterion for entry-level positions. As

indicated in section 4.3.1 above, the Norwegian Air Force used 12 years of education as a

formal criterion for admission to their ab initio training programme. In research conducted by

Van Eeden et al. (2001), it was indicated that school performance in Mathematics (r = 0.25; p

< 0.01), Science (r = 0.28; p < 0.01 and English (r = 0.33; p < 0.01) was the best predictor of

average first-year performance in science courses.

A study by Shochet (1998) showed that while there was a statistically significant correlation

between Matric results and university success for educationally advantaged students - with the

Matric results explaining 30% of the variance of student’s average result - a similar conclusion

could not be drawn for academically disadvantaged students. However, Zaaiman (1998)

concluded that the South African Matric results are not valid or reliable because of differences

in the scholastic education of students.

4.4 DATA COLLECTION AND PROCESSING

According to Mouton and Marais (1990), the step of data collection in the research design

poses a huge challenge for the social science researcher because of the rational, historical

and normative characteristics of humans. They (1990) further stated that it is essentially a

requirement that the application of a valid measuring instrument to different groups under

different sets of circumstances should lead to similar observations.

The purpose of gathering data is to ensure optimal administration of the collection methods, as

well as comprehensiveness and accuracy of the raw data in respect of the independent and

dependent variables.

Huysamen (2000, p. 146) asserted that “ it is common cause that students from schools

formerly falling under the Department of Education and Training (DET) have been exposed to

141

schooling system which was inferior compared to that attended by their white counterparts”.

Hence the Matric results of students from these schools, particularly black students, are bound

to correlate more poorly with tertiary academic performance (Huysamen, 2000). In his study to

investigate whether university marks were a better predictor of marks in subsequent years

than Matric marks, he operationalised the Matric performance by assigning values of 8 to 2 to

symbols A to G obtained for subjects passed at the higher grade (HG), values of 6 to 0 to the

corresponding symbols for standard grade (SG) and summing these values to obtain a Matric

symbol point total (MSPT).

In the current research, a similar procedure was used to convert the Matric symbols to values,

which enabled the researcher to conduct statistical analysis on this variable.

4.4.1 Data collection for the independent variables

A psychometrist and a test administrator administered the tests at the various centres

countrywide. The test administrator functioned as an assistant during the testing process, in

distributing and collecting the test material, answering questions and ensuring that examples

were completed and properly understood for each test battery. This approach was used for

both phases 1 and 2 of the selection process.

Prior to the final selection process, all the applicants who were found suitable in phase2 of the

psychometric tests were required to perform aircraft manoeuvres in an aircraft simulator to

measure their psychomotor capabilities based on motor coordination of arms/hands and legs,

spatial skills in a three-dimensional direction, multitasking by processing information from three

instruments at the same time, strategic information processing by using the memory, attention

to detail, information gathering, evaluation and logical processing and mathematical operations

using a compass. The applicants were given practice tasks before the actual test. It is

pertinent to mention that these results were not considered in the current study.

4.4.2 Data collection for the dependent variable

The final flying school training results were provided by the flying school which had conducted

both the theoretical and practical flying tests results. The theoretical results were used in the

current research.

142

4.5 DATA ANALYSIS

The purpose of the data analysis is to test the hypothesis and the general aims of the

research, as set out in section 1.3.

4.5.1 Introduction

According to Kerlinger, (1986, p. 175), "Statistics is the theory and method of analysing

quantitative data obtained from a sample of observations in order to study and compare

sources of variance of phenomena, to help make decisions to accept or reject hypothesized

relations between the phenomena, and to aid in making reliable inferences from empirical

observations."

In the current research, statistical analysis was used to test the hypotheses that the

psychometric test results during phases 1 and 2 correlated with performance in terms of the

actual flying school final examination results in a cadet pilot training programme.

All statistical analyses in the present study were computed using the SPSS statistical package

for Windows version 11.1 (SPSS, 2001).

There are two major components of the discipline of statistics - descriptive and inferential

statistics. Also of importance are the concepts of statistical significance, practical effect sizes,

reliability and validity, all of which will be explained in this section.

4.5.2 Descriptive statistics

The purpose of descriptive statistical analysis is to provide an overview of the data that were

collected for the research. The tables in chapter 5 reflect the sample size (n) in the initial

selection process based on the paper screening in terms of race and gender as well as the

mean and standard deviation for each variable.

143

According to Cohen (2008), descriptive statistics is concerned with summarising a given set of

measurements, whereas inferential statistics involves generalising beyond the given data to

some larger potential set of measurement.

Descriptive statistics were calculated for each variable in the study. In the case of categorical

data, for example, this involved the calculation of the frequency distribution of the responses

to each categorical scaled question. A frequency distribution shows in absolute or relative

(percentage) terms, how often the different values of a variable are found among the

respondents. In the current study, biographical data such as race and gender are examples of

variables that are presented using frequency distribution data.

The measures of central or average tendency are the mean and variance. Although they both

measure optimised sets of scores, they do so in different ways. According to Kerlinger and Lee

(2000), the mean expresses the general level and is representative of the level of a group’s

characteristics or performance.

The variance is a measure of dispersion of the set of scores in that it indicates how much the

scores are spread out. In other words, the variance describes the extent to which the scores

differ from one another. For descriptive purposes, the square root of the variance is used. This

is referred to as the standard deviation (SD) (Kerlinger & Lee, 2000).

In the cases where scales were used in the various tests, the descriptive statistics that were

most appropriate were means and standard deviations.

4.5.3 Inferential statistics

The inferential statistical techniques used in the study are discussed below in more detail.

Inferential statistics is the area of statistics described by Howitt and Cramer (2000) as using

sample scores to make general statements or draw general conclusions that apply well

beyond the sample being used in the research.

The basic inferential techniques used in this study were analysis of variance (ANOVA),

correlation and regression analysis.

144

4.5.3.1 ANOVA

In general, the purpose of ANOVA is to test the statistical significance of differences between

means (Clayton, 1984). In the present study, the four race groups were compared with respect

to their mean scores on the various tests used in phases 1 and 2. For this purpose, the

ANOVA procedure (Kerlinger, 1986) was used to examine the effect of the various variables,

for instance, race and gender on the successful completion of the cadet pilot training

programme.

According to Kerlinger (1986), unfortunately tests of statistical significance such as the t- and

F-tests do not indicate the magnitude or strength of relations. An F-test, if significant, simply

says that a difference exists. When the F-test in the ANOVA proves significant at, say, the 0.05

level, the researcher knows that there are statistically significant differences between the

groups (e.g. race groups) as far as the dependent variable (which would be one of the

psychometric tests performed) is concerned. However, this is often merely saying that the

groups are different with respect to their mean scores, and it is still not known which pairs of

groups (of the possible pairings of groups) are different in respect of their mean scores.

For this purpose, when the overall F-test of the ANOVA is significant, a so-called “post-hoc

test” procedure is applied in order to test which pairwise group differences are significant. The

procedure used in the present study is the Scheffé test (Glass & Stanley, 1984).

Furthermore, according to Kerlinger (1986), the Scheffé test, if used with discretion, is a

general method that can be applied to all comparisons of means after an analysis of variance.

The main point is that post-hoc comparisons and tests of means can be conducted mainly for

exploratory and interpretative purposes, as is the case in the current study.

4.5.3.2 Correlation analysis

The specific aim of this study was to evaluate the predictive validity of the psychological test

battery used in the selection of cadet pilots as a predictor of success in the training

programme. In other words, the aim was to determine if there is a relationship between the

results of the psychological tests used and the final flying school results.

145

Another specific aim of this research was to investigate whether the English Matric symbol,

literacy and the ABET levels correlated with performance in the final flying school examination

results for the different race and gender groups.

According to Shaughnessy (2000), a correlation exists when two different measures vary

together - that is, when the scores on one variable co-vary with scores on another variable.

The correlation coefficient measures the strength of the linear relationship between variables.

The value of the correlation coefficient ranges from –1 (for a perfect negative correlation) to +1

for a perfect positive correlation. One of the most frequently used measures of correlation is

the Pearson product-moment correlation (r).

A positive correlation between two variables means that as the scores on one of the variables

increase, the scores on the other variable also tend to increase. If the correlation is negative,

however, say, -0.5, then the higher a person’s score on one variable, the lower his/her score is

likely to be on the other variable.

In the current study, the effect size indexes of about 0.10 will be considered small,

correlations of 0.30 moderately large and correlations of about 0.50 large (Cohen, 1992).

These parameters are set to guide the interpretation and drawing of inferences from the

results of the research.

Miles and Banyard (2007) asserted that a correlation is both a descriptive and an inferential

statistic.

The correlations in the current research were calculated to determine the relationships

between the psychological tests utilised in both phases 1 and 2 and the final flying school

results.

4.5.3.3 Regression analysis

According to Howitt and Cramer (2003), regression, like correlation coefficient, describes

important features of a scattergram relating two variables, but allows the researcher to make

predictions. In the current study, the stepwise regression analysis was conducted to examine

146

the separate effects of two or more predictors on the criteria (Tredoux & Durrheim, 2002). The

effects of the various psychological tests on performance on the different modules of the flying

school were investigated.

.

4.5.4 Statistical techniques used

The statistical techniques used in this research are presented in this section. The following two

options were considered:

(1) Variables measured on an interval scale. In such cases, the Pearson product

moment correlation coefficient (Kerlinger, 1986) is calculated as a measure of the

linear relationship between the different variables.

In the current study, the hypotheses were mostly about positive correlations between

tests and performance in the cadet pilot training programme. Tests scores were all

interval measurements and the Pearson correlation coefficient was used. According to

Howitt and Cramer (2003, p. 52), “it is more convenient to have the main features of

the scattergram expressed as a single numerical index – the correlation coefficient”.

The correlation coefficient helps to present the research finding more concisely,

particularly when several variables are involved and they are considered a basic

descriptive statistic. Therefore, owing to the number of variables in the current

research, it was deemed appropriate to calculate the Pearson correlation coefficient to

determine the relationships between the various dependent and independent variables.

According to Howitt and Cramer (2003), the correlation coefficient consists of the following two

parts:

a positive or negative sign

any numerical value in the range of 0.00 to 1.00

Based on the numerical value and the sample size, the statistical significance determines the

interpretation of results in terms of the testing of the hypothesis in research.

(2) One variable categorical and another measured on an interval scale. In such

cases, the ANOVA technique was used to identify whether significant differences

147

between the categorical groups existed and a post-hoc Scheffe test was used to

identify the exact groups that differ from one another.

Examples in the current study refer to differences between the scores of different race groups

(categorical variable) on the test scores (interval variables).

4.5.5 Level of statistical significance

According to Field (2005), there are two possibilities in the real world (the actual population):

(1) In reality, there is an effect in the population.

(2) There is no effect in the population.

Breakwell, Hammond and Fife-Schaw (2000) explained that statistical significance tests begin

with the supposition that the null hypothesis is true. However, the null hypothesis always

assumes that there is not a difference in means. More specifically, the null hypothesis states

that there is no difference between, say, two mean scores or no real relationship between

variables in the population. The alternative hypothesis, however, states that there is a real

difference or relationship.

The probability of obtaining the observed data if the null hypothesis is true is therefore

calculated during most inferential tests (such as ANOVA and correlation analysis). If the

probability is small (say, 0.05), it is unlikely that the null hypothesis is true, and one could

therefore conclude that the null hypothesis is false.

There is always a chance that the researcher might be wrong in his/her decision. According to

Tredoux and Durrheim (2002), using samples to estimate population parameters always yields

uncertain results. If the researcher rejects the null hypothesis and concludes that the

population means are not equal, when in fact they are in the real population, then a type I

error was made. A type I error therefore occurs when a researcher rejects a null hypothesis,

that is, H0, when it is true. However, a type II error occurs when H0 is false, but the conclusion

made is that H0, is true. Generally, a type II error is represented by β (beta).

Conventionally, most researchers use the significance levels of 0.05 and 0.01. These are small

values because the researcher wishes to be sure before he/she comes forward with a

148

significant result. The intention is to limit the risk of committing a type I error. One would rather

run the risk of missing a significant result than erroneously producing one.

In the current study, a significance level of 0.05 was used. This view was supported by

Kerlinger (1986) in this statement that the 0.05 level is still used by researchers because it is

considered to be a reasonably sound gamble. It is neither too high nor too low for most social

scientific research. However, the researcher will not focus on statistical significance only but

also note the actual size of the results such as the size of the correlations.

4.5.6 Practical effect size

Leong and Austin (2006) emphasised the need to extend research beyond the

significant/nonsignificant dichotomy. They also cited other authors such as Cohen (1992) in

support of this. These authors feel that it is important not to limit all research to studies whose

results will hinge on a significant test and that a valuable contribution to theory should not be

hindered by the pervading acceptance of the p < 0.05 requirement.

While there are various methods of calculating the effect size, in the current study, the GLM

procedure of the statistical package SPSS was used to calculate the ANOVA, which also

calculates and presents the so-called “partial squared eta values” which were used as an

indication of the practical effect size (Pierce, Block & Aguinis, 2004). According to them (2004,

p. 918), partial eta squared is described as the “proportion of the total variation attributable to

the factor, partialling out (excluding) other factors from the total non-error variation” and is

normally higher than eta squared.

In the current research, partial eta squared was calculated for each race group for the

psychological tests utilised in phase 1. The results will be provided in tables 5.9 to 5.15.

The generally accepted regression benchmark for effect size comes from (Cohen, 1992): 0.20

is a minimal solution (but significant in social science research); 0.50 indicates medium effect;

and anything equal to or greater than 0.80 is considered to be a large effect size (Cohen,

1992).

The regression analysis, helps to make predictions about the relationship between the

variables, was calculated. The aim of the research was to also determine what psychological

149

tests would predict successful completion of the cadet pilot programme using the final flying

school results as criteria.

However, Cohen (1992) acknowledged the danger of using terms like ”small”, ”medium” and

”large” out of context. Glass and Stanley (1994) were particularly critical of this approach,

arguing that the effectiveness of a particular intervention can only be interpreted in relation to

other interventions that seek to produce the same effect.

4.6 RESEARCH HYPOTHESES

The specific aims of the current research were discussed in chapter 1 (see section 1.4). The

research hypotheses are discussed below.

Mouton and Marais (1990) discussed the obvious use of hypotheses in quantitatively oriented

research, and indicated the following requirements for a hypothesis:

It should be stated explicitly, at least in the form of a research question.

It should be formulated beforehand.

It should be possible to reject it.

Quantitative researchers are far more concerned with ensuring that the hypotheses have been

formulated before embarking on the investigation (Mouton & Marais, 1990). Hence in the

current study, the hypotheses set out below, were formulated to cover the objectives of the

empirical study. The hypotheses were aligned to the quantitative-oriented research as outlined

above.

4.6.1 Hypotheses regarding the predictive validity of each psychological test in phase 1

of the selection for performance in a cadet pilot training programme

The following hypotheses were formulated on the predictive validity of the psychological tests

for performance in the flying school final examination results in a cadet pilot training

programme:

150

H11: Raven’s Progressive Matrices scores correlate statistically significantly and positively

with the flying school final examination results.

H21: The Blox Test scores correlate statistically significantly and positively with the flying

school final examination results.

H31: The scores for reading comprehension in the Intermediate Test Battery correlate

statistically significantly and positively with the flying school final examination results.

H41: The arithmetic (parts 1 and 2) subtest scores correlate statistically significantly and

positively with the flying school final examination results.

H51: The literacy level scores of the ELSA scores correlate statistically significantly and

positively with the flying school final examination results.

H61: The ABET level scores of the ELSA correlate statistically significantly and positively

with the flying school final examination results.

H71: The Matric English symbol scores correlate statistically significantly and positively

with the flying school final examination results.

4.6.2 Hypotheses regarding the predictive validity of each psychological subtest in

phase 2 of the selection for performance in a cadet pilot training programme

Hypotheses were formulated on the predictive validity of each psychological test in phase 2 of

the selection for performance in the flying school final examination results in a cadet pilot

training programme.

H81: The comprehension subtest of the WAIS scores correlates statistically significantly

and positively with the flying school final examination results.

H91: The arithmetic subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

H101: The digit forward subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

H111: The digit backward subtest of the WAIS scores correlates statistically significantly

and positively with the flying school final examination results.

H121: The digit combine subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

151

H131: The similarity subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

H141: The picture completion subtest of the WAIS scores correlates statistically

significantly and positively with the flying school final examination results,

H151: The object assembly subtest of the WAIS scores correlates statistically significantly

and positively with the flying school final examination results.

H161: The block subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

H171: The digit ’90 subtest of the WAIS scores correlates statistically significantly and

positively with the flying school final examination results.

H181: The picture Arrangement subtest scores of the WAIS correlate statistically

significantly and positively with the flying school final examination results.

4.7 CHAPTER SUMMARY

This chapter dealt with the empirical study. This involved a discussion of the population and

sample, the measuring instruments used, the objectives of the study in terms of the predictor

and criterion variables and the gathering and statistical processing of the data for the various

techniques that were used to determine the predictive validity of each of the psychological

tests.

The chapter concluded by formulating the research hypotheses relating to the validity of each

of the psychological tests used to predict performance in a cadet pilot training programme.

Chapter 5 will discuss the research findings and interpret the results of the data analysis.

152

CHAPTER 5

THE RESEARCH RESULTS 5.1 INTRODUCTION In this chapter the results of the empirical study are discussed. The objective of the

research was to determine which of the variables that were used significantly predict

successful completion of the cadet pilot training programme.

The results below first indicate the performance in psychological tests during the initial

phase (phase 1), split by race and gender. In the next section then the performance in

the tests used in phase 2 are discussed, with a look at differences between the race

groups.

In the last section, the relationships and performance in the first and second phase tests

are discussed. Correlations between the first and second phase tests as well as the

performance of the respondents in the cadet pilot training programme are presented.

Regression analysis is used to predict performance in the modules of the cadet training

programme.

5.2 RESULTS OF THE PSYCHOMETRIC TESTS IN PHASE 1: COMPARISON

BETWEEN THE RACE AND GENDER GROUPS

In this section, the descriptive statistics for the sample on the independent variable in

phase 1 are presented in table 5.1.

Secondly, the scores for the males and females and for the different race groups were

examined and an ANOVA procedure performed to determine whether there were any

statistically significant differences between the race and gender groups in the test scores

in phase 1.

It was previously reported that the total sample consisted of 2 238 applicants who had

met the minimum criteria in terms of the information on the application forms, namely

Matric certificate, Science, Maths and English as their subjects, and an indication of their

153

race and gender and South African citizenship. The race and gender of the applicants

were noted but not considered during the selection process except for comparison of

subgroups.

Table 5.1 shows the number of applicants per race group. Blacks had the highest

number of applicants (n = 1 334), followed by whites. In terms of the purpose of the

cadet pilot training programme, which was to improve diversity in the workforce, it can be

inferred that overall, EE requirements were achieved in that the previously

disadvantaged population groups were well represented. However, the distribution per

race group was disproportionate in that there were significantly more white applicants

than coloureds and Asians. The mean scores indicate that blacks scored lower in all the

tests, in comparison with the other race groups, while the white applicants scored the

highest.

In Wheeler’s (1993) study on apprentices in a mining industry, he found a significant

difference in means between the black and white groups on all the predictors, that is, the

mental alertness test (t = 4.39, p < 0.001) and the Blox Test (t = 7.68,p < 0.001), which

led the researcher to conclude that the mean scores indicated that on average, the black

apprentices tended to achieve lower scores on the predictors. In Wheeler’s (1993)

research, the blacks generally scored between a half to one standard deviation lower

than their white counterparts in these tests.

A similar trend was evident in the current study (table 5.1), where the blacks achieved

almost similar values, with the exception of Raven’s Test where they scored three

standard deviations above their counterparts, while the coloureds and blacks scored

between a half to one standard deviation lower than the Asians and whites, in the Blox

Text, and the reading comprehension and arithmetic 1 tests.

In a study conducted by Freedman (2000, p. 83) using Raven’s SPM, in the mining

industry, it was reported that “given that the maximum score obtained on Raven’s SPM

is 60, a score of 40, 45 on this test for a given respondent may be expected to vary

between 35, 06 and 45.84”. Freedman (2000) concluded that Raven’s SPM remains a

reasonable predictor of literacy and numeracy performance and should be retained as a

screening test for selection in a training programme.

154

5.2.1 Results of the tests per race group on the independent variables

Table 5.1

Descriptive statistics for the psychometric tests in phase 1

TESTS RACE Mean Std. deviation

N

Raven’s

Black 44.4 9.368 1334 Coloured 51.1 5.341 122 Asian 52.4 4.530 258 White 55.1 3.436 524 Total 48.2 9.010 2 238

Blox

Black 24.4 6.700 1334 Coloured 29.8 6.171 122 Asian 30.8 5.797 258 White 35.0 4.783 524 Total 27.9 7.644 2 238

Reading comprehension

Black 7.2 3.284 1334

Coloured 11.9 3.112 122

Asian 12.9 2.863 258

White 14.3 2.911 524

Total 9.8 4.485 2 238

Arith- metic 1

Black 14.6 4.534 1334 Coloured 18.5 3.975 122 Asian 20.4 3.590 258 White 20.5 3.764 524 Total 16.9 5.076 2 238

Arithmetic 2

Black 8.4 2.394 1 334

Coloured 11.0 2.443 122

Asian 12.5 2.234 258

White 12.7 2.061 524

Total 10.1 3.044 2 238

155

Figure 5.1 below plots these mean scores to present a graphic image of the scores.

Figure 5.1 shows that the whites performed better in all the tests, followed by the Asians,

the coloureds and the blacks.

Mean score per Race group per first fase psychometric

test

0.00

20.00

40.00

60.00

Mean

sco

re

RAV

Blox

RC

AR1

AR2

RAV 44.40 51.11 52.37 55.10

Blox 24.39 29.77 30.77 35.04

RC 7.16 11.89 12.94 14.28

AR1 14.57 18.54 20.40 20.53

AR2 8.45 10.98 12.49 12.74

Black Coloured Asian White

Figure 5.1. Mean score per race group for the psychometric tests in phase 1 A possible reason why the blacks scored lower than the other groups could be that most

of the tests were conducted in English. However, the Raven and Blox Tests are

regarded as nonverbal tests. It could be argued that no test is entirely nonverbal

primarily because the test instructions are normally explained verbally.

According to Nell (1999, p. 12), “it is by no means clear that home language

administration is the method of choice, since translating a test from English into the

subjects’ home language may block access to terms and concepts they have acquired

through the medium of the English language”.

Table 5.2 below indicates which differences are statistically significant as well as the

practical effect size (partial eta). If the partial eta is 0.01, a small effect size is indicated,

156

while 0.06 indicates a medium effect siza and 0.14 or higher a strong effect size (Cohen,

1992).

Table 5.2 ANOVA comparison of race groups in each of the tests

Dependent variable F Sig. Partial eta squared

Raven’s 280.6 .000 .274

Blox 399.6 .000 .349

Reading comprehension 772.6 .000 .509

Arithematic 1 327.4 .000 .305

Arithematic 2 556.1 .000 .428

The larger the sample size, the more likely it is that statistically significant results can be

obtained, and indeed, the race groups differed statistically significantly in all the tests.

The partial eta values, however, indicated large effect sizes (values close to 0.14 or

larger) and it may therefore be concluded that practically significant differences do exist.

The post-hoc tests were computed to determine between which race groups differences

exist. Tables 5.3 to 5.7 indicate these post-hoc tests.

In table 5.3, the coloureds’ and Asians’ performance in Raven’s test was more or less

the same, with the whites performing moderately better then the coloureds and Asians.

The blacks performed below the other race groups in this test.

157

Table 5.3 Scheffé subsets of race group on Raven’s Test

RACE

N

Subset

1 2 3

Blacks 1334 44.4033

Coloureds 122 51.1066

Asians 258 52.3721

Whites 524

55.0992

Rushton et al. (2003) reported on a study conducted in 2000 at the University of

Witwatersrand in Johannesburg, where three separate studies were conducted using the

standard Raven’s Progressive Matrices. They found that the 198 African first-year

engineering students had an average IQ of 97, in contrast to the white students in the

three studies who had IQs ranging from 105 to 111; the Indian students had intermediate

IQs ranging from 102 to 106.

Table 5.4 indicates the performance on the Blox Test, with the coloureds and Asians

again on par (i.e. their performance was fairly similar), with the whites performing better

than the other race groups. Again the blacks were at the bottom end of the scale.

Table 5.4 Scheffé subsets of race group on the Blox Test

RACE

N

Subset 1 2 3

Blacks 1 334 24.3928

Coloureds 122 29.7705

Asians 258 30.7674

Whites 524 35.0363

Table 5.5 indicates the results of the four race groups based on their performance in the

reading comprehension test. The differences between these subgroups show that the

whites performed better, followed by the Asians, the coloureds and the blacks.

158

Table 5.5 Scheffé subsets of race group on the reading comprehension test

RACE

N

Subset

1 2 3 4

Blacks 1334 7.1589 Coloureds 122 11.8934

Asians 258 12.9419

Whites 524

14.2824

Tables 5.6 and 5.7, which show the participants’ performance in arithmetic 1 and 2

respectively, indicate that the Asian and whites subgroups had similar results in these

tests. The coloured subgroup performed better than the blacks in both tests.

Table 5.6 Scheffé subsets of race group in arithmetic 1

RACE

N

Subset

1 2 3

Blacks 1334 14.5750

Coloureds 122 18.5410

Asians 258 20.3992

Whites 524

20.5305

Table 5.7 Scheffé subsets of race group in Arithmetic 2

RACE

N

Subset

1 2 3

Blacks 1334 8.4453

Coloureds 122 10.9836

Asians 258 12.4922

Whites 524

12.7366

5.2.1.1 Summary

According to Howitt and Cramer (2000), if there is a significant difference between all the

groups, then one would have four separate culture groups each with only one group. In

other words, the tables above indicate that there were differences between the groups.

159

It appears that differences do exist between all the race groups in terms of the reading

comprehension test. In terms of the Raven and Blox Tests, the coloureds and the Asians

were generally on a par, whereas their scores differed from those of the blacks and

whites. In arithmetic 1 and 2, the white and Asian respondents performed similarly, but

their scores differed from those of the blacks and coloureds, who in turn, also differed

from one another.

Six studies in sub-Saharan Africa supported Spearman’s (1927) hypothesis that

black/white IQ differences are mainly on g, the general factor of intelligence (Rushton et

al. (2003). These authors reported on analysed item data by Rushton in 2000 from 40

000 high school students in South Africa with regard to the Standard Progressive

Matrices published by Owen (1992) and found that the four-way African-coloured-

Indian-white mean differences were all on g.

5.2.2 Results for the tests by gender (within race)

The differences between the males and females in these tests are compared below for

each of the race groups. First gender is compared for the black race group. The mean

scores in table 5.8 below indicate that the males scored higher in the Raven and Blox

Tests, with the means equal to 44.49 and 24.81 respectively, while the females scored

higher in the reading comprehension test, with a mean of 8.06. The scores in the

arithmetic tests were virtually identical.

160

5.2.2.1 Blacks Table 5.8 Descriptive statistics for gender among blacks on the independent variables

Variable Gender Mean Std. deviation N

Raven’s

Male 44.49 9.344 1 171

Female 43.81 9.550 163

Total 44.40 9.368 1 334

Blox

Male 24.81 6.684 1 171

Female 21.37 6.019 163

Total 24.39 6.700 1 334

Reading Comprehension

Male 7.03 3.217 1 171

Female 8.06 3.612 163

Total 7.16 3.284 1 334

Arithmetic 1

Male 14.57 4.571 1 171

Female 14.59 4.267 163

Total 14.58 4.534 1 334

Arithmetic 2

Male 8.47 2.391 1 171

Female 8.28 2.418 163

Total 8.45 2.394 1 334

To test whether the differences observed in the table above are statistically significant an

ANOVA was performed, with the results indicated in table 5.9 below. Although the

females performed significantly lower on the Blox Test (p = 0.000), the practical effect

size was small (close to 0.01). In the case of reading comprehension, the females

actually performed better than the males, but again the practical effect size was small.

161

Table 5.9 ANOVA comparison of the black gender groups

Variable F Sig. Partial eta squared

Raven’s .745 .388 .001

Blox 38.9 .000 .028

Reading comprehension 14.1 .000 .011

Arithmetic 1 .002 .966 .000

Arithmetic 2 .861 .354 .001

Although statistically significant differences were found for the Blox Test and reading

comprehension results, the effect size results indicate that these differences were not

practically significant, showing small effect size (eta = 0.028, p < 0.000 and eta = 0.011,

p < 0000) respectively. While there were no real differences between the results of the

black males and females in these tests, no black females managed to advance to phase

2.

Considering the mean scores for the black males and females (see table 5.8), and the

fact that the females performed significantly higher in reading comprehension, while the

males performed significantly higher in the Blox, it was surprising that no black females

were able to advance to phase 2. This outcome merits further investigation.

162

5.2.2.2 Coloureds Table 5.10 Descriptive statistics per gender group for the coloureds for the independent variables

Variable Gender Mean Std. deviation N

Raven’s

Male 51.58 5.200 97

Female 49.28 5.594 25

Total 51.11 5.341 122

Blox

Male 30.58 6.178 97

Female 26.64 5.147 25

Total 29.77 6.171 122

Reading comprehension

Male 11.84 3.155 97

Female 12.12 2.991 25

Total 11.89 3.112

122

Arithmetic 1

Male 18.39 4.112 97

Female 19.12 3.407 25

Total 18.54 3.975 122

Arithmetic 2

Male 11.03 2.400 97

Female 10.80 2.646 25

Total 10.98 2.443 122

The results of the descriptive statistics for the coloureds on the independent variable in

phase 1 are indicated in table 5.10.

The table indicates that the males scored higher in the Raven’s, Blox and Arithmetic 2

tests, whereas the females achieved higher scores in the reading comprehension and

arithmetic 1 test.

163

Table 5.11 ANOVA comparison of the coloured gender groups

Variable F Sig. Partial eta squared

Raven’s 3.761 .055 .030

Blox 8.600 .004 .067

Reading comprehension

.166 .685 .001

Arithmetic 1 .665 .416 .006

Arithmetic 2 .176 .675 .001

It is only in the Blox Test that the coloured males performed statistically significantly

better than the females. The practical effect size was of medium strength (eta = 0.067,

p<0.005).

164

5.2.2.3 Asians Table 5.12 Descriptive statistics per Asian gender group for the independent variables

Variable Gender Mean Std. deviation N

Raven’s

Male 52.58 4.349 212

Female 51.29 5.221 45

Total 52.35 4.530

257

Blox

Male 31.53 5.441 212

Female 27.11 6.132 45

Total 30.75 5.805 257

Reading comprehension

Male 12.85 2.739 212

Female 13.44 3.354 45

Total 12.96 2.858 257

Arithmetic 1

Male 20.66 3.586 212

Female 19.36 3.304 45

Total 20.43 3.567 257

Arithmetic 2

Male 12.75 2.187 212

Female 11.33 2.056 45

Total 12.51 2.228 257

The mean scores indicate that the results of the females and males in this group

were similar, with the exception of the Blox Test, where the males scored better

than the females.

165

Table 5.13 ANOVA comparison of the Asian gender groups

Variable F Sig. Partial eta squared

Raven’s

3.041 .082 .012

Blox 23.375 .000 .084

Reading comprehension 1.590 .209 .006

Arithmetic 1 5.009 .026 .019

Arithmetic 2 16.000 .000 .059

Once again, the Asian females performed significantly lower on the Blox Test compared

with the Asian males’ medium to strong practical effect size (eta = 0.084, p = 0,000). The

Asian females also achieved lower scores in arithmetic 2 (eta = 0. 059, p = 0.000) with

medium practical effect size.

166

5.2.2.4 Whites Table 5.14 Descriptive statistics per white gender group for the independent variables

Variable Gender Mean Std. deviation N

Raven’s

Male 55.2 3.41 402

Female 54.8 3.50 122

Total 55.1 3.44 524

Blox

Male 35.9 4.41 402

Female 32.2 4.90 122

Total 35.0 4.78 524

Reading comprehension

Male 14.1 2.90 402

Female 14.9 2.88 122

Total 14.3 2.91 524

Arithmetic 1

Male 20.6 3.82 402

Female 20.5 3.58 122

Total 20.5 3.76 524

Arithmetic 2

Male 12.8 2.09 402

Female 12.5 1.94 122

Total 12.7 2.06 524

Table 5.14 indicates that the scores on the independent variables were markedly similar

in all the tests, with the exception of the Blox Test, where the males had higher scores

than the females ( mean = 35.9 and SD = 4.41).

167

Table 5.15 ANOVA comparison of the White gender groups

Variable F Sig. Partial eta squared

Raven’s 1.315 .252 .003

Blox 62.242 .000 .107

Reading comprehension 6.338 .012 .012

Arithmetic 1 .057 .811 .000

Arithmetic 2 1.957 .162 .004

The white females performed significantly lower on the Blox Test (p = 0.000). The

practical effect size was medium to large (eta = 0.107). This indicates that the white

females outperformed the white males in the reading comprehension test (with p =

0.012). However, the practical effect size was small (eta = 0.012).

5.2.2.5 Summary of descriptive results

The black males and females showed statistically significant and practical significance

effect size differences between the tests, and there were generally few large differences

(large eta) between the gender groups in the different race groups.

It would appear, however, that the male respondents generally outperformed the

females in the Blox Test, but the opposite applied in the the reading comprehension test.

5.2.3 RELATIONSHIPS BETWEEN PSYCHOMETRIC TESTS (PHASE 1)

To determine whether there was a relationship between the results of the respondents in

the different tests, a correlation matrix was performed between all the five tests in first

phase. Table 5.16 depicts the correlation matrix.

168

Table 5.16 Correlation matrix of the tests in phase 1

Variable Raven’s Blox

Reading comprehen sion

Arithmetic 1

Arithmetic2

Correlation

Raven’s 1.000 .648* .554* .510* .585*

Blox .648* 1.000 .556* .479* .568* Reading Comprehension

.554* .556* 1.000 .569* .643*

Arithmetic 1 .510* .479* .569* 1.000 .701* Arithmetic 2 .585* .568* .643* .701* 1.000

*p < 0.05

All significance values (p-values) were higher than 0.05, indicating statistically highly

significant results. It can be concluded that all the tests correlates statistically highly

significantly with one another. The Pearson r (correlation coefficient) showed a large

effect size, indicating positive strong relationships.

All the correlations were positive, which indicates that the higher a respondent’s scored

in one test, the higher he/she was likely to score in the other tests as well, and vice

versa.

5.3 THE RESULTS OF THE PSYCHOMETRIC SUBTESTS DURING PHASE 2:

COMPARISON OF RACE GROUPS

The mean scores on each subtest of the WAIS (tests performed during phase 2) as well

as Matric performance in English (symbol), literacy and ABET levels are presented

below for each of the race groups.

169

Table 5.17 Descriptive statistics for the test results for phase 2, per race group

RACE Mean Std. deviation N

LITERACY

Black 3.05 1.715 19

Coloured 4.93 0.997 14

Asian 4.89 1.323 18

White 5.57 0.778 35

Total 4.77 1.516 86

ABET

Black 2.32 0.820 19

Coloured 3.07 0.730 14

Asian 3.11 0.758 18

White 3.57 0.698 35

Total 3.12 0.873 86

English Matric symbol

Black 2.00 0.816 19

Coloured 2.79 0.699 14

Asian 3.06 1.259 18

White 3.40 1.063 35

Total 2.92 1.129 86

Comprehension

Black 12.37 2.733 19

Coloured 13.64 2.437 14

Asian 12.17 1.855 18

White 14.11 2.349 35

Total 13.24 2.478 86

Arithmetic

Black 6.89 2.401 19

Coloured 6.50 1.951 14

Asian 8.00 3.087 18

White 8.40 2.316 35

Total 7.67 2.541 86

Digit forward

Black 7.11 1.100 19

Coloured 7.36 0.633 14

Asian 7.94 0.802 18

White 7.63 0.942 35

Total 7.53 0.942 86

Digit backward

Black 5.00 1.764 19

Coloured 5.57 1.697 14

Asian 6.17 1.295 18

White 6.23 1.477 35

Total 5.84 1.600 86

Digit combined

Black 12.11 2.233 19

Coloured 12.79 1.424 14

Asian 14.11 1.641 18

White 13.86 2.102 35

Total 13.35 2.074 86

170

RACE Mean Std. deviation N

Similarities

Black 17.37 2.948 19

Coloured 18.71 1.978 14

Asian 17.78 1.865 18

White 17.71 2.527 35

Total 17.81 2.423 86

Picture completion

Black 12.95 1.471 19

Coloured 13.86 0.949 14

Asian 13.83 1.295 18

White 14.06 0.906 35

Total 13.73 1.202 86

Object assembly

Black 17.53 3.502 19

Coloured 18.50 1.605 14

Asian 17.00 2.828 18

White 20.31 2.958 35

Total 18.71 3.177 86

Block design

Black 29.74 5.645 19

Coloured 30.71 3.688 14

Asian 30.50 5.904 18

White 36.40 3.790 35

Total 32.77 5.551 86

Digit 90

Black 57.42 8.002 19

Coloured 59.71 9.352 14

Asian 59.50 10.107 18

White 61.03 10.340 35

Total 59.70 9.593 86

Picture arrangement

Black 12.74 3.588 19

Coloured 15.07 3.605 14

Asian 14.83 2.895 18

White 14.91 3.373 35

Total 14.44 3.432 86

171

Table 5.17 indicates the differences between the different race groups. The standard

deviation of the blacks was between half and two below the total group standard

deviation in all the subtests.

An ANOVA was performed to establish which differences were of importance from a

statistical and practical effect size perspective.

The results of this analysis are presented below in table 5.18.

Table 5.18 Tests of between-subjects’ effects per test between race groups

Variable F Sig. Partial eta squared

LITERACY 18.608 .000* .405

ABET 11.717 .000* .300

Matric English symbol 8.075 .000* .228

Comprehension 3.834 .013* .123

Arithmetic 2.812 .044* .093

Digit forward 2.919 .039* .097

Digit backward 3.014 .035* .099

Digit combined 4.668 .005* .146

Similarities .876 .457 .031

Picture completion 4.035 .010* .129

Object assembly 6.754 .000* .198

Block design 11.759 .000* .301

Digit 90 .575 .633 .021

Picture arrangement 2.097 .107 .071

*p < 0.05;

172

There were statistically significant differences between the race groups in most of the

tests, with the partial eta squared values indicating medium to very large differences

between the race groups. The exceptions were the tests “Similarity”, “Digit 90” and

“Picture arrangement”, although even for these tests, a slightly larger than small effect

size was obtained.

The post-hoc Scheffé test indicated significant differences for the tests items with bullets

below.

Literacy (eta = 0.405). The post-hoc Scheffé results identified the homogeneous

subsets indicated in table 5.19.

Table 5.19 depicts the correlations between the race groups in terms of the scores for

the literacy independent variable.

Table 5.19 Scheffé test results for level of literacy

RACE

N

Subset

1 2

Black 19 3.0526

Asian 18 4.8889

Coloured 14 4.9286

White 35 5.5714

The results show that the level of literacy of the blacks was generally lower than that of

the other three race groups. The remaining three race groups did not differ statistically

significantly from one another.

ABET levels (eta squared = 0.300). The post-hoc Scheffé results identified the

homogeneous subsets shown in table 5.20. In other words, the scores for the

coloureds, Asians and whites were similar in terms of ABET levels, whilst those

for the blacks were below the other race groups.

173

Table 5.20 Scheffé results for the ABET levels

RACE

N

Subset

1 2

Black 19 2.3158

Coloured 14 3.0714

Asian 18 3.1111

White 35 3.5714

The results indicate that the blacks’ ABET levels were generally lower than those of the

other three race groups. The other three race groups did not differ statistically

significantly from one another.

Matric English symbol (eta squared = 0.228). The post-hoc Scheffé results

identified the homogeneous subsets indicated in table 5.21 below.

Table 5.21 Scheffé results for the Matric English symbol

RACE

N

Subset

1 2

Black 19 2.00

Coloured 14 2.79

Asian 18 3.06

White 35 3.40

The blacks performed at a lower level in Matric English, while statistically, the coloureds,

Asians and whites did perform significantly differently from one another.

This could be symptomatic of the different education systems that were in place for the

various race groups prior to 1994.

Digits combined (eta squared = 0.146)

174

Table 5.22 Scheffé results for digits combined

RACE

N

Subset

1 2

Black 19 12.1053

Coloured 14 12.7857

Asian 18 14.1111

White 35 13.8571

The significant difference here was between the blacks (who scored lowest for digits

combined) and the Asians who scored highest. While the scores for the blacks and

coloureds were similar, they were, however, statistically significantly different.

Picture completion (eta squared = 0.129)

Table 5.23 Scheffé results for picture completion

RACE

N

Subset

1 2

Black 19 12.9474

Coloured 14 13.8571 Asian 18 13.8333 White 35 14.0571

The blacks scored significantly lower than the whites, coloureds and Asians, whose

scores did not differ statistically significantly from one another. The coloureds and

Asians scored the highest and second highest respectively in this test, with a small

difference between the two subgroups.

Object assembly (eta squared = 0.198)

175

Table 5.24 Scheffé results for object assembly

RACE

N

Subset

1 2

Black 19 17.5263

Coloured 14 18.5000

Asian 18 17.0000

White 35 20.3143

The Asians scored lowest in this test and significantly lower than the whites. Although

other differences were not statistically significant, it is interesting to note that the

coloureds scored second highest, followed by the blacks who, in turn, outperformed the

Asians.

Block design (eta squared = 0.198) Table 5.25 Scheffé results for block design

RACE

N

Subset

1 2

Black Coloured

19 14

29.7368 30.7143

Asian 18 30.5000

White 35 36.4000

The whites scored significantly higher in this test than the other race groups. However,

the other race groups did not differ statistically significantly from one another.

5.4 SUMMARY OF THE COMPARISON OF THE SUBGROUP RESULTS

The white respondents generally performed better in most of the tests except for digits.

The Asians did best in this test. The differences, between these two groups, however,

were not statistically significant.

176

Although the black respondents generally performed poorest on most of the tests, there

were exceptions. For instance, they outperformed the Asian respondents in the object

assembly test (although not significantly so). Furthermore, they also did not perform

significantly worse than the coloured and white respondents in the digit test or

significantly worse than the coloured and Asian respondents in the picture completion

and block design tests.

Nell (1999) reported on preliminary data collected by Unisa’s Health Psychology unit for

157 black South Africans with less than 12 years’ education. Those in long-standing

employment in a sophisticated environment, scored between about one and two

standard deviations below the US WAIS–R norms using the digit symbol, block design,

digit span (forward and backward) and arithmetic subtests.

Nell (1999) also reported on research conducted on university undergraduates which

showed that these undergraduates were one standard deviation below the US age

norms on the first three subtests of the verbal scale and on block design; nonsignificantly

below the norm on digit span and similarities; and more than a standard deviation below

the norm on picture completion, picture arrangement and object assembly.

5.5 CORRELATION OF PHASE 1 AND PHASE 2 RESULTS WITH

PERFORMANCE IN THE CADET PILOT TRAINING PROGRAMME

It is pertinent to mention at this stage that it is the final results achieved at the flying

school and obtaining the Commercial Pilot Licence that actually determines the final

successful outcome. The following questions can be asked about each respondent’s

performance in the cadet pilot training programme:

Do the candidates’ test scores obtained in the first phase of selection show

statistically significant positive correlations with their performance in the cadet

pilot training programme?

177

Do the additional tests scores that became available in the second phase show

statistically significant positive correlations with their performance in the cadet

pilot training programme?

5.5.1 Correlation of phase 1 results with performance in the cadet pilot training

programme

One would normally conduct regression analyses with the performance in cadet pilot

training programme as a dependent variable. However, it is clear that the dynamics differ

from one race group to the next. If regression analyses were to be performed, they

would therefore have to be performed for each race group separately. Because the

sample sizes were too small for multiple regression analyses based on race groups, it

was decided to perform regression analyses and correlate the tests of the first and

second phases of the selection process with the performance in the cadet pilot training

programme on the total sample of 28 respondents. This sample represents all the

candidates who were finally selected for the actual training programme at the flying

school.

178

Table 5.26 Correlations between the phase 1 test results and the final flying school results

Variable/modules Raven’s Blox Test Reading compre-hension

Arithmetic 1

Arithmetic 2

Meteorology Correlation P-value

0.376(*) 0.049

- 0.062 0.755

0.078 0.695

0.005 0.980

0.108 0.583

Flight planning Correlation P-value

- 0.119 0.547

- 0.089 0.651

0.274 0.158

0.207 0.290

0.161 0.415

Radio Correlation P-value

0.054 0.783

- 0.024 0.902

- 0.065 0.741

- 0.015 0.941

0.077 0.696

Navigation Correlation P-value

0.273 0.159

0.018 0.927

- 0.073 0.711

0.292 0.132

0.287 0.138

Instrumentation Correlation P-value

0.420(*) 0.026

0.135 0.493

0.272 0.162

0.409(*) 0.031

0.372 0.051

Air law & procedures

Correlation P-value

- 0.271 0.163

-0.102 0.604

- 0.205 0.295

- 0.234 0.231

- 0.134 0.496

Human performance

Correlation P-value

-0.141 0.475

0.023 0.908

0.041 0.835

0.096 0.629

0.164 0.404

Aircraft technical Correlation P-value

0.059 0.765

0.106 0.592

- 0.009 0.964

0.100 0.614

0.050 0.802

* Correlation is significant at the 0.05 level (2-tailed).

179

Table 5.26 Correlations between the phase 1 test results and the final flying school results (continued)

Variable/modules ELSA ABET English language symbol

Matric symbol

Meteorology Correlation P-value

0.212 0.299

0.224 0.272

0.331 0.098

0.231 0.256

Flight planning Correlation P-value

0.234 0.250

0.129 0.531

0.208 0.307

0.133 0.517

Radio Correlation P-value

0.170 0.407

0.233 0.251

0.343 0.087

0.431(*) 0.028

Navigation Correlation P-value

0.448 (*) 0.022

0.473(*) 0.015

0.295 0.144

0.205 0.315

Instrumentation Correlation P-value

0.297 0.141

0.229 0.260

0.406(*) 0.040

0.282 0.163

Air law & procedures

Correlation P-value

0.035 0.867

0.037 0.859

- 0.186 0.363

- 0.077 0.709

Human performance

Correlation P-value

0.269 0.185

0.267 0.188

-0.092 0.655

0.090 0.663

Aircraft technical Correlation

P-value

0.317 0.114

0.267 0.188

0.377 0.057

0.304 0.131

* Correlation is significant at the 0.05 level (2-tailed).

180

Table 5.26 indicates that statistically significant positive correlations exist between

Raven’s (RPM) and two measures of the cadet pilot training modules, namely

meteorology and instrumentation. One other significant correlation was evident between

instrumentation and arithmetic 1 in the phase 1 psychometric results.

The following is a summary of the significant results of the correlation analysis:

The correlation between the RPM and the meteorology module of the flying school

results was statistically significant (r = 0.376; p < 0.05).

Also, the RPM showed a positive correlation with the instrumentation module of the

flying school results, which was statistically significant (r = 0.420; p < 0.05).

The arithmetic 1 subtest indicated a high correlation with the instrumentation module,

which was statistically significant (r = 0.409; p<0.05).

The results show a correlation of a medium to large effect size. However, in terms of

statistical significance, two correlations reached statistical significance at the 0.05 level,

namely the correlation between Raven’s test and meteorology and instrumentation and

between arithmetic 1 and instrumentation.

The following are possible explanations for the correlations or the lack thereof:

Raven’s Progressive Matrices test measures an individual’s general intelligence -

hence an individual who identifies a theme in terms of a response to the test items

obtains a high score. The results of the RPM indicated a positive correlation with

meteorology, radio, navigation, instrumentation and the aircraft technical modules of

the flying school (dependent variable), except for the flight planning, air law and

procedures and human performance modules which reflect a negative correlation.

Only two of the correlations were significant and the negative correlations were not

significant.

While there were not many significant correlations, it is pertinent to state that certain

abilities, such as identifying patterns/themes/strategies and learning within time

181

constraints, as tested by the RPM, remain critical pilot skills because pilots are

required to make quick decisions under stressful conditions to ensure flight safety

Flotman (2002).

Flotman (2002, p. 61) noted that “the Advanced Raven’s test provides an indication of

a person’s intellectual efficiency (e.g. to make quick, accurate judgments’ under

stressful conditions) which is critical during flying process”. The researcher found a

positive correlation between this test and the ground school phase results of the

candidates.

The Blox Test reflected a negative correlation with most of the modules, but where

the correlations were positive, they were insignificant. The measure of the Blox Test

is spatial ability, which includes visualisation, spatial relations and spatial orientation,

which are deemed to be critical skills for a pilot. The negative correlation in this study

may have been as a result of the individual candidates not obtaining the required

pass mark during the first sitting of the pilot training examination.

Flotman (2000, p. 61) asserted that “despite the Blox test’s poor correlation with the

Ground School Phase results, it remains an important component of the selection

battery in the sense that it measures one of the critical abilities a potential pilot should

have, namely, spatial relations and orientation”. The above research recommended

that the test should be retained until a more comprehensive study is undertaken.

The reading comprehension test of the intermediate battery measures verbal

ability and clerical speed and accuracy. The poor correlation with the flying school

modules could be the result of the technical language used during the training

programme compared with the language used in the test. Hence the performance of

the candidates during the selection phases may have indicated potential success in

the training, although the training school results do not appear to have confirmed this.

The correlations between arithmetic 1 and 2 and the flying school results were

not significant, with the exception of arithmetic 1 and the instrumentation module,

which were statistically significant. If one considers the fact that the items in the tests

are verbal statements of computational problems, these may not be similar to

182

calculations required when flying an aircraft. This could be an indication that a study

needs to be conducted to determine the nature and type of numerical abilities

required for a pilot, and whether these differ from the general numerical abilities

based on community norms as discussed in chapter 3.

5.5.2 Correlations between the phase 2 test results and performance in the cadet

pilot training programme

It was decided not to perform regression analyses for each race group because the

sample sizes were too small. Instead the correlations and regression of phase 2 tests

with performance in cadet pilot training programme based on the total sample were

investigated.

Burke, Hobson and Linsky (1997) noted a trend of declining validity over time for several

pilot aptitude measures and a discernible trend for smaller validation sample sizes in

more recent decades.

183

Table 5.27 Correlations between the phase 2 test results and the final flying school results

Variable/modules ELSA ABET English language symbol

Matric symbol

Meteorology Correlation P-value

0.212 0.299

0.224 0.272

0.331 0.098

0.231 0.256

Flight planning Correlation P-value

0.234 0.250

0.129 0.531

0.208 0.307

0.133 0.517

Radio Correlation P-value

0.170 0.407

0.233 0.251

0.343 0.087

0.431* 0.028

Navigation Correlation P-- value

0.448 * 0.022

0.473* 0.015

0.295 0.144

0.205 0.315

Instrumentation Correlation P-value

0.297 0.141

0.229 0.260

0.406* 0.040

0.282 0.163

Air law & procedures Correlation P-value

0.035 0.867

0.037 0.859

- 0.186 0.363

- 0.077 0.709

Human performance Correlation P-value

0.269 0.185

0.267 0.188

-0.092 0.655

0.090 0.663

Aircraft technical Correlation

P-value

0.317 0.114

0.267 0.188

0.377 0.057

0.304 0.131

*p < 0.05

184

Table 5.27 Correlations between the phase 2 test results and the final flying school results (continued)

Variable/modules Comprehension Arithmetic Digit forward

Digit backward

Digit combined

Similarities

Meteorology Correlation P-value

0.292 0.148

0.324 0.106

0.262 0.196

0.150 0.466

0.243 0.231

-0.148 0.471

Flight planning Correlation P-value

- 0.037 0.856

- 0.123 0.549

0.129 0.531

0.068 0.740

0.115 0.575

-0.085 0.681

Radio Correlation P-value

0.286 0.157

0.204 0.317

- 0.045 0.828

0.099 0.632

0.062 0.764

-0.210 0.302

Navigation Correlation P-value

0.053 0.796

-0.004 0.985

- 0.098 0.633

0.010 0.962

-0.036 0.860

0.297 0.140

Instrumentation Correlation P-value

- 0.004 0.985

- 0.059 0.775

0.085 0.680

0.104 0.612

0.125 0.542

-0.031 0.880

Air law & procedures

Correlation P-value

0.026 0.901

- 0.112 0.587

- 0.288 0.153

- 0.187 0.360

- 0.286 0.156

-0.160 0.435

Human performance

Correlation P-value

0.289 0.152

0.086 0.675

- 0.043 0.836

-0.258 0.203

-0.235 0.249

0.124 0.547

Aircraft technical Correlation

P-value

0.396* 0.045

-0.249 0.221

- 0.196 0.336

-0.196 0.336

-0.181 0.377

-0.324 0.106

*p < 0.05

185

Table 5.27 Correlations between the phase 2 test results and the final flying school results (continued)

Variable/modules Picture completion

Object assembly

Block design

Digit 90 Picture arrange-

ment

Meteorology Correlation P-value

0.033 0.872

0.145 0.480

0.001 0.995

0.024 0.909

-0.030 0.885

Flight planning Correlation P-value

0.034 0.867

- 0.133 0.518

0.032 0.875

-0.236 0.246

-0.105 0.611

Radio Correlation P-value

-0.393(*) 0.047

- 0.284 0.159

- 0.168 0.411

- 0.229 0.260

-0.013 0.951

Navigation Correlation P-value

0.013 0.950

- 0.103 0.617

- 0.108 0.600

-0.092 0.260

0.148 0.470

Instrumentation Correlation P-value

0.005 0.982

0.121 0.555

0.029 0.889

- 0.128 0.534

0.018 0.932

Air law & procedures

Correlation P-value

- 0.147 0.473

- 0.340 0.089

- 0.310 0.123

- 0.394* 0.047

- 0.224 0.270

Human performance

Correlation P-value

-0.128 0.534

- 0.423* 0.032

-0.313 0.119

0.009 0.966

-0.087 0.672

Aircraft technical Correlation P-value

-0.164 0.424

0.035 0.864

- 0.154 0.454

0.055 0.454

-0.114 0.580

*p < 0.05

186

Table 5.27 indicates that there were generally few statistically significant correlations

between the phase 2 tests and the performance of cadet pilots during training. This

could be the result of restriction of range which could have affected the correlation sizes

because of preselection and a small sample size.

According to Burke et al. (1997), sample sizes in pilot validation are important because

the occupation of aviation pilot attracts a highly self-selected group of applicants as a

career option, and those entering training typically represent a small percentage of the

most able of these applicants. Hence these high degrees of selection tend to bias

estimates of validity downwards. Thorndike (in Burke et al., 1997) observed that

validities tend to underestimate the true validities of pilot selection measures simply

because the full range of ability is not present in all validation samples on account of

range restriction.

Hunter and Burke (in Burke et al., 1997) reported with regard to the nature of the

criterion widely used in pilot selection research, that the dichotomous split of training

success into pass versus fail tends to dominate the pilot selection literature, a factor that

would also substantially reduce the observed validity estimates.

A similar comment is made by Stauffer and Ree (1996), who postulated that often in

assessing pilot selection systems, continuous criteria may not be available for all

trainees, particularly those who fail and those who leave before completing all the

phases of instruction that contribute to continuous criteria. This scenario was

experienced during the cadet training programme on which the results of this study were

based, where at least two of the candidates left the programme for personal reasons.

The results based on the phase 2 psychometric battery are summarised below.

The correlation between the comprehension and aircraft technical module was

statistically significant (r = 0.396; p < 0.05). The comprehension test measures

the respondent’s ability to deal with abstract social conventions, rules and

expressions, while the aircraft technical modules include material on the different

aircraft types, and so on, which are considered abstract.

187

The negative correlations between picture completion and the radio module,

object assembly and the human performance module as well as digit 90 and air

law and procedures results, were statistically significant (r = - 0.393, - 0.423 and

– 0.394; p < 0.05) respectively. Picture completion and object assembly involve

the ability to deal with perceived visual details. Radio focuses on communication

with control centres and other air space users, while human performance has to

do with, say, workload and stress and their impact on pilot performance. Object

assembly measures the ability to a form visual concept quickly, which is a critical

skill for a pilot. Hence the statistically significant correlation with human

performance reflects this relationship.

Although the results of literacy and ABET showed positive correlations with air

law and procedures, which were not statistically significant, this could be an

indication of the content of the subject/module and the language, as opposed to

the language of the test items of the ELSA, in which the language is more South

African.

The correlation between literacy level, as measured by the ELSA, with the

navigation module was statistically significant (r = 0.448; p<0.05). The ELSA

quantifies a person’s English competency input and trainability levels, and the

results reflect positive correlations with the some of the flying school modules.

The relationship between the literacy levels of a cadet pilot and his/her

understanding of signals on the dials in the aircraft was reflected in this

statistically significant correlation.

There was a statistically significant correlation between the ABET levels and

navigation (r = 0.473; p < 0.05). Also, the ABET level (i.e. academic level) of the

cadet pilot was reflected in the relationship between the predictor and the

criterion.

The Matric English language symbol indicated a positive correlation with both the

instrumentation and aircraft technical modules of the flying school final

examination results, which were statistically significant (r = 0.406; p < 0.05 and r

= 0.377; p < 0.05 respectively). Again, the academic English level of the cadet

188

pilot and the understanding of the various instruments used to manoeuvre an

aircraft and dials reflected a positive relationship.

The correlation between the Matric symbol and the radio module was positive

and statistically significant (r = 431; p < 0.05). The Matric symbol as reflected by

the academic certification of the cadet pilot indicated a relationship with

communication via radio in an aircraft.

In conclusion it should be noted that the results seems to indicate poor correlations in

phases 1 and 2 as reported in tables 5.26 and 5.27. This could have been caused by the

job description used to identify the psychological tests and qualities to be assessed

during the recruitment and selection process being based on the competencies of

qualified pilots. Furthermore, the cadet pilots may not have had prior exposure to and

experience in aviation and performance during the selection process may thus have

been influenced by other (unknown) factors.

Burke et al. (1997) supported the view of Ackerman (1989) and Fitts and Posner (1967)

to improve prediction in pilot selection, by following a skills acquisition model, in order to

obtain incremental benefits to add tests of information-processing abilities to account for

increasing dominance of cognitive ability in the later stages of pilot training and in coping

with the information demands of contemporary avionics systems in actual operations.

Sommer et al. (2004) pointed out that in many fields; there are low correlation

coefficients between test results and the chosen criterion variable. There are a variety of

reasons for this. They stated that based on the test results obtained in the selection

procedures, individuals who are already excluded are not available for further evaluation

analysis. This selection results in a restriction of the range in the predictor variable and

thus a decrease in the correlation coefficient (Sommer et al., 2004).

5.5.3 STEPWISE REGRESSION ANALYSIS

In section 5.5.2 it was stated that multiple regression was performed on the total sample

and the psychological tests in both phases 1 and 2 in order to determine which tests are

better predictors of performance in the cadet training programme. The stepwise

189

regression analysis method was used because it is the most parsimonious model

(Brace, Kemp & Snelgar, 2009). Furthermore, the total number of psychological tests

including the subtests used during the selection of the cadet pilots, leaned towards using

the stepwise regression method because it would ensure ending up with the smallest

possible set of predictor variables included in the model.

The regression analysis was conducted on the basis of the statistically significant

relationship between the psychological tests and the final cadet flying school results.

The variables included as part of the model in the stepwise regression analysis were as

follows:

(1) Dependent variables

Meteorology

Navigation

Radio

Air law and procedures

Human performance

Aircraft technical

(2) Independent variables

Raven’s Progressive Matrices test score

The Blox Test score

ABET levels

the Matric symbol

the digit 90 subtest score

the object assembly subtest score

the comprehension subtest score

According to Howitt and Cramer (2000), the primary use of regression is that it allows

the researcher to make predictions. Stepwise regression was used to ensure that the

smallest possible set of predictor variables was obtained taking into account the small

sample size. This view is supported by Brace et al. (2009), in their statement that the

advantage of the stepwise method is that it always results in the most economical

model. This suggests that if a researcher wishes to determine the minimum number of

190

variables that he/she needs to measure in order to predict the criteria variable, the

stepwise method would be a better option. Brace et al. (2009) recommended that if one

does not have a strong theoretical framework, then it is probably safest to use the

“enter”.

Owing to the multicolinearity nature of the variable, a variable either being entered on its

own or with other variables in the stepwise procedure will result in the percentage of the

variable explained being different.

The results of the stepwise regression are presented in tables 5.28 to 5.33.

Table 5.28 below indicates the regression results for the meteorology module in the

cadet training programme.

Table 5.28

Predicting performance in meteorology

Model R R square Adjusted R square

Std. error of the estimate

1 .470(a) .221 .189 6.297769

2 .672(b) .452 .404 5.397591

a. Predictors: (constant), RPM b. Predictors: (constant), RPM, Blox Test *p < 0.05

191

Coefficient

Model Unstandardised coefficients

Standardised coefficients

t Sig.

B Std. error Beta

1 (Constant) 45.147 15.075 2.995 .006

RAVEN’S .723 .277 .470 2.609 .015

2 (Constant) 31.833 13.611 2.339 .028

RAVEN’S 1.523 .350 .989 4.350 .000

BLOX -.884 .284 -.707 -3.110 .005

a. Dependent variable: meteor *p < 0.05

The tables above indicate that by including only RPM only 19% of variance could be

explained. However, adding the Blox to the model, 40% (R2 change = 0.404) of variance

in the dependent variable could be explained. This indicates that both the RPM and the

Blox can be used to predict performance in meteorology. The beta values of both RPM

and the Blox are presented in the table above.

Adjusted R square = 0.404

The significant model, using the stepwise method, has the following significant variables:

Predictor variable Beta p

Raven’s 0.989 p < 0.0005

Blox - 0.707 p = 0.005

192

Table 5.29

Predicting performance in navigation

Model R R

square

Adjusted

R square

Std. error of the

estimate

1 .473(a) .224 .192 8.076926

a. Predictors: (constant), ABET

*p < 0.05

Model

Unstandardised coefficients

Standardised coefficients

t Sig.

B

Std. error

Beta

1 (Constant) 76.159 4.192 18.168 .000

ABET 3.104 1.326 .431 2.341 .028

a. Dependent variable: navigation *p < 0.05

The ABET level results were the only variable that had any relationship with the

dependent variable flying school module (navigation). The ABET levels explain

19.2% of the variance in the dependent variable.

Olea and Ree (in Carretta, et al., 1996), in their research on predicting situation

awareness, demonstrated the predictive power of general cognitive ability for

pilot and navigator criteria.

Adjusted R square = 0.224;

The significant model, using the stepwise method, has the following significant

variables:

Predictor variable Beta p

ABET 0.431 p = 0.028

193

Table 5.30

Predicting performance in radio

Model R R

square Adjusted R square

Std. error of the estimate

1 .431(a) .186 .152 8.154893

a. Predictors: (constant), Matric symbol *p < 0.05

Model Unstandardised coefficients

Standardised coefficients

t Sig.

B

Std. error

Beta

1 (Constant) 76.159 4.192 18.168 .000 MATRICSB 3.104 1.326 .431 2.341 .028

a. Dependent variable: radio *p < 0.05

The Matric symbol obtained by the candidates, who were finally selected for the training

programme, had a significant predictive relationship with the dependent variable flying

school module (radio). The Matric symbol explains 15.2% of the variance in the

dependent variable.

Adjusted R square = 0.152;

The significant model, using the stepwise method, had the following significant variables:

Predictor variable Beta p

Matric symbol 0.431 p = 0.028

Table5.31

Predicting performance in air law and procedures

Model R R square

Adjusted R square

Std. error of the estimate

1 .394(a) .155 .120 5.678382

a. Predictors: (constant), digit 90 *p < 0.05

194

Model Unstandardised coefficients

Standardised coefficients

t Sig.

B Std. error Beta

1 (Constant) 97.915 7.973 12.281 .000 DIGIT90 -.273 .130 -.394 -2.099 .047

a Dependent variable: air law *p< 0.05

The table indicates that digit 90 was the variable that had the strongest linear

relationship with the independent variable (air law and procedures) of all the

independent variables. Digit 90 explained 12% of the variance in the dependent variable.

Adjusted R squared = 0.120

The significant model, using the stepwise method, had the following significant variables:

Predictor variable Beta p

Digit 90 -0.394 p = 0.047

Table 5.32

Predicting performance in human performance

Model R R square

Adjusted R square

Std. error of the

estimate

1 .423(a) .179 .144 5.258605

a. Predictors: (constant), object assembly

*p < 0.05

Model Unstandardised coefficients

Standardised coefficients

t Sig.

B

Std. error

Beta

1 (Constant) 106.571 8.916 11.953 .000

OBJECT -1.146 .502 -.423 -2.284 .032

a. Dependent variable: human performance *p < 0.05

195

The object assembly subtest had the best relationship with the human performance of

the flying school of all the independent variables, but the adjusted R square of 0.144

indicated that object assembly explained only 14.4% of the dependent

variable.<,Adjusted R square = 0.144

The significant model, using the stepwise method, had the following significant variables:

Predictor variable Beta p

Object assembly -0.423 p = 0.032

Table 5.33

Predicting performance in aircraft technical

Model R R

square

Adjusted

R square

Std. error of

the estimate

1 .396(a) .157 .122 6.563980

(a) Predictors: (constant), comprehension

*p < 0.05

Model Unstandardised coefficients Standardised coefficients

t Sig.

B Std. error

Beta

1 (Constant) 70.083 6.601 10.617 .000 COMPR 1.023 .484 .396 2.114 .045

(a) Dependent variable: aircraft *p < 0.05

The linear relationship between with dependent variable (i.e. aircraft technical) and the

independent variable (i.e. comprehension subtest) was significant. The comprehension

subtest indicated 12.2% of variance in the dependent variable. In other words, a small

part of the variance of the dependent variable is explained by the adjusted R square.

Adjusted R square = 0.122

The significant model, using the stepwise method, had the following significant variables:

196

Predictor variable Beta p

Comprehension 0.396 p = 0.045

5.6 INTEGRATION OF RESULTS

The results of the research suggest that the psychometric batteries utilised in phases 1

and 2 of the selection process for the cadet pilot training programme do have some

benefits and value in identifying potential candidates who could successfully complete

the training programme. The conclusions of the research are as follows:

5.6.1 Descriptive statistics and group comparison

The whites in phase 1 of the selection processes performed better, followed by the

Asians, the Coloureds and the blacks (figure 5.1) in all the psychometric batteries.

The blacks in phase 2 of the selection process generally achieved lower scores in

the literacy levels compared with other race groups, which did not differ significantly

from one another, as measured by the ELSA subset 1 = 3.0526 for the blacks (see

table 5.19).

The blacks generally achieved lowers scores than the other race groups, which did

not differ significantly from one another, on the ABET levels as measured by the

ELSA subset 1 = 2.3158 for the blacks (see table 5.20).

The blacks and coloureds had similar results in Matric English. in subset 1, blacks =

2.00 and the coloureds = 2.79. However, the coloureds, Asians and whites did not

perform significantly differently from one another (see table 5.21).

5.6.2 Correlation analysis

The correlation between the RPM and the meteorology module of the flying school

results was statistically significant (r = 0.376; p < 0.05) (see table 5.26).

The literacy levels and the ABET indicated a statistically significant correlation with

the navigation module (r = 0.448 and 0.473; p < 0.05 respectively) (see table 5.26).

The Matric symbol, as indicated on the applicants’ application forms also had a

statistically significant correlation with the radio module (r = 0.431; p < 0.05) (see

table 5.26).

197

The correlation between comprehension and the aircraft technical module was

significant (r = 0.396; p < 0.05) (see table 5.27).

Despite the negative correlation between picture completion and the radio module,

object assembly and the human performance module as well as digit 90 and the air

law and procedures results, the correlation was significant (r = - 0.393, - 0.423 and –

0.394; p < 0.05) respectively (see table 5.27).

5.6.3 Stepwise regression analysis

The beta weighting indicated that overall, the RPM, Blox, Matric English symbol,

ABET levels and reading comprehension were good predictors for some of the

modules of the flying school results (see tables 5.28 to 5.33).

Some of the possible contributions of these results were discussed earlier. This forms

the foundation on which to test the hypotheses formulated in chapter 4.

5.7 TESTING OF THE RESEARCH HYPOTHESES

In this section, each hypothesis is either rejected or accepted on the basis of the results

of the empirical research as reported in this chapter. According to Anastasi and Urbina

(1997), a significance level of less than 0.01 or 0.05 indicates that the null hypothesis

should be rejected.

5.7.1 Hypotheses testing regarding the psychological tests

Table 5.34 represents the results of this research in terms of the hypotheses that were

stated in sections 4.6.1 and 4.6.2 in chapter four.

198

Table 5.34 Hypothesis testing of the results

Original hypotheses

Results Reasons/discussion

H11: The RPM scores correlate statistically

significantly and positively with the flying

school final examination results.

H10 is rejected. The results indicate that the RPM is valid for

the prediction of performance in the flying

school final examination results, namely the

instrumentation and meteorology modules in

cadet training programme.

H21: The Blox Test scores correlate

statistically significantly and positively with the

flying school final examination results,

H20 is not rejected. The results seem to indicate that the Blox Test

is not valid for the prediction of performance in

the flying school final examination results in

the cadet training programme.

H31: Reading comprehension in the

Intermediate Test Battery scores correlates

statistically significantly and positively with the

flying school final examination results.

H30 is not rejected. The results seem to suggest that there is no

relationship between the test and performance

in the flying school final examination results in

cadet training programme.

199

Table 5.34 Hypothesis testing of the results (continued)

Original hypotheses

Results Reasons/discussion

H41: The arithmetic (1 and 2) subtests scores

correlate statistically significantly and positively

with the flying school final examination results.

H40 is not rejected. The results indicate that the arithmetic 1 & 2

subtests are not valid for the prediction of

performance in the flying school final

examination results in the cadet training

programme. However, there is some indication

that arithmetic 1 does have a correlation with

the instrumentation module in the flying school

results.

H51: The literacy level scores of the ELSA

correlate statistically significantly and positively

with the flying school final examination results.

H50 is rejected. The results indicate that the literacy level

scores of the ELSA are valid for predicting

performance in the navigation module in the

flying school final examination results in the

cadet pilot training programme.

H61: The ABET level scores of the ELSA

correlate statistically significantly and positively

with the flying school final examination results.

H60 is rejected. The results indicate that the ABET level scores

of the ELSA are valid for predicting

performance in the navigation module in the

flying school final examination results in the

cadet pilot training programme.

200

Table 5.34 Hypothesis testing of the results (continued)

Original hypotheses

Results Reasons/discussion

H71: The Matric English symbol scores

correlate statistically significantly and positively

with the flying school final examination results.

H70 is rejected. The results indicated that the Matriculation

English symbol is valid for predicting

performance in the instrumentation and aircraft

technical modules in the flying school final

examination results in the cadet pilot training

programme.

H81: The comprehension subtest of the

WAIS scores correlates statistically

significantly and positively with the flying

school final examination results.

H80 is rejected. The results indicate that the comprehension

subtest of the WAIS is valid for predicting

performance in the aircraft module in the flying

school final examination results in the cadet

pilot training programme.

H91: The arithmetic subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results.

H90 is not rejected. The results indicate that the arithmetic subtest

of the WAIS is not valid for predicting

performance in the flying school final

examination results in the cadet pilot training

programme.

201

Table 5.34 Hypothesis testing of the results (continued)

Original hypotheses

Results Reasons/discussion

H101: The digit forward subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results.

H100 is not rejected. The results indicate that the digit forward

subtest of the WAIS is not valid for predicting

performance in the flying school final

examination results in the cadet pilot training

programme.

H111: The digit backward subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results

H110 is rejected. The results indicate that the digit backward

subtest of the WAIS is valid for predicting

performance in the flying school final

examination results in the cadet pilot training

programme.

H121: The digit combined subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results.

H120 is not rejected. The results indicate that the digit span subtest

of the WAIS is not valid for predicting

performance in the flying school final

examination results in the cadet pilot training

programme.

202

Table 5.34 Hypothesis testing of the results (continued)

Original hypotheses

Results Reasons/discussion

H131: The similarities subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results.

H130 is not rejected. The results indicate that the similarities subtest of the WAIS is not valid for predicting performance in the flying school final examination results in the cadet pilot training programme.

H141: The picture completion subtest of the

WAIS scores correlates statistically

significantly and positively with the flying

school final examination results.

H140 is rejected. The results indicate that the picture completion subtest of the WAIS is valid for predicting performance in the radio module in the cadet pilot training programme.

H151: The object assembly subtest of the

WAIS scores correlates statistically

significantly and positively with the flying

school final examination results.

H150 is rejected. The results indicate that the object assembly

subtest of the WAIS is valid for predicting

performance in the human performance

module in the flying school final examination

results in the cadet pilot training programme.

203

Table 5.34 Hypothesis testing of the results (continued)

Original hypotheses

Results Reasons/discussion

H161: The block design subtest of the WAIS

scores correlates statistically significantly and

positively with the flying school final

examination results.

H160 is not rejected. The results indicate that the block design

subtest of the WAIS is not valid for predicting

performance in the flying school final

examination results in the cadet pilot training

programme.

H171: The digit 90 subtest of the WAIS scores

correlates statistically significantly and

positively with the flying school final

examination results.

H170 is rejected. The results indicate that the digit 90 subtest of the WAIS is valid for predicting performance in the air law and procedures module in the flying school final examination results in the cadet pilot training programme.

H181: The picture arrangement subtest of the

WAIS scores correlates statistically

significantly and positively with the flying

school final examination results.

H180 is not rejected. The results indicate that the picture

arrangement subtest of the WAIS is not valid

for predicting performance in the flying school

final examination results in the cadet pilot

training programme.

204

5.7.2 Discussion

The results obtained highlight areas in which the psychological tests currently used in

the selection of cadet pilot are valid in predicting performance in the training programme

for specific modules. The RPM, which provides an indication of intelligence in the

context of the current research, was not consistently valid in all the modules of the flying

school.

The Blox Test results obtained indicated a poor correlation. However, the regression

analysis provided a different picture, which suggests that the Blox Test used in

conjunction with the RPM would provide a better prediction.

Further indications from the results obtained relate to the literacy and ABET levels of the

candidate as well as the English Matric symbol, which showed a statistically significant

correlation with some of the modules of the flying school, particularly those relating to

radio, navigation, instrumentation and aircraft technical.

It is interesting to note that the psychological tests, the Blox Test, reading

comprehension and arithmetic 2 in phase 1, did not indicate any statistically significant

correlations with the modules of the flying school final examination results. In phase 2 of

the selection process, the following subtests of the WAIS did not show statistically

significant correlations with the all the modules of the flying school final examination

results: arithmetic, digit forward and backwards, digit span, similarities, block design and

picture arrangement. A possible reason for this could be that the sample size is too small

or that general intelligence is not necessarily as critical as special abilities/aptitude for

performance as a pilot.

In studies conducted locally on military pilots, it was concluded that the predictive validity

of general cognitive ability (g) and specific abilities (sn) appears to differ (De Kock &

Schlechter, 2009). In their meta-analysis of predictors of pilot success, Hunter and Burke

(1994) found that general intelligence was not generalisable across studies as a

predictor; at most it had an influence moderated by other variables. However, Ree and

Carretta (1996) concluded that general cognitive ability has consistently been shown to

205

predict pilot training success, showing average statistically significant correlations of

0.33.

Damon (1996) reported that some evidence has been found that information processing

capability is an important indicator of fluid intelligence, thus predicting pilot success. A

South African study found that pilots could be differentiated from non-pilots on the basis

of the rate of information processing (Barkhuizen, Schepers & Coetzee, 2002; De Kock

& Schlechter, 2009). According to De Kock and Schlechter (2009), it can be argued that

fluid intelligence and information processing capability are two factors of intelligence that

drive transfer and automatisation of learning during flight training tasks.

As mentioned in chapter 2, according De Kock and Schlechter (2009), clearly the debate

on the role of intelligence and aptitude in the prediction of pilot training success is still

active and can be interpreted as an attestation of its dominance in pilot selection

batteries.

De Kock and Schlechter (2009), reported on a study by Burke, Hobson and Linsky

(1997), which found that psychomotor tests predicted pilot success and that their validity

generalised across samples. The above authors used validity generalisation analysis

(VGA) with three samples from different national air forces, with a large sample (N = 1

760). According to De Kock and Schlechter (2009), as a follow-up to these findings,

various other studies reported that measures of psychomotor abilities were able to

increase the predictive validity of a battery already measuring (g) (De Kock & Schlechter

2009; Ree & Carretta, 1996).

De Kock and Schlechter (2009) concluded that this construct might prove useful in future

pilot selection batteries.

According to Ree and Carretta (1996), some studies did in fact report that certain

aspects of personality had incremental predictive validity in traditional batteries, for

instance, attitude to risk. De Kock and Schlechter (2009) reported on a study by Carretta

(2000), who that found that a measure of conscientiousness increased the multiple

correlation coefficient of a battery measuring general mental ability, from 0.51 to 0.60.

206

In another study, pilot trainees completed a personality inventory measuring five

dimensions deemed to be associated with flight training performance. On completion of

the training, three of the measures were in fact significantly related to training outcome,

namely hostility, self-confidence and value flexibility. Disappointingly, incremental validity

analysis did not indicate that the inventory could enhance a selection model already

containing traditional aptitude scores (De Kock & Schlechter 2009).

Campbell, Castaneda and Pulos (2010, p. 97), in their meta-analysis studies using

personality constructs to predict military aviation outcomes, noted that of the 26 studies

reporting the effect of personality, excluding biographical inventories, “all but two studies

[utilized] military samples, suggesting the use of personality assessment in aviation

training contexts is even less applied in civil aviation”. Because of this comment, the

civilian samples were omitted in their study. This view further indicates the need for more

research in civilian/commercial airlines, as it can be argued that the personality profile of

a commercial pilot would differ significantly from that of a military pilot.

In conclusion, the results of these studies underscore the view that not only intelligence,

but also other domains, such as personality, information processing, and so forth, may

play a role in predicting flying training outcomes. It is clear that the current battery may

be deficient in the sense that it does not include personality, as suggested by a literature

study by De Kock and Schlechter (2009), which was also the case in the current

research. Personality was not considered in the present study.

These studies confirm that a thorough study should be undertaken to identify the critical

competencies of a pilot, which will ensure that appropriate and relevant psychological

instruments are used in the selection process. Campbell et al. (2010) concluded that the

implication of their study is that the differential relationship between specific personality

traits and aviation training success should move beyond applications using monolithic

“personality” as a predictor into a more theoretical modelling of the interaction between

specific personality constructs.

Overall, the aim of this research, namely to determine the predictive validity of the

current psychological test battery in order to measure successful completion of the cadet

training programme, has been achieved.

207

5.8 CHAPTER SUMMARY

The results of the empirical study were presented and discussed in this chapter. The

results indicated that there is a concern about the weak and/or small correlation between

the subtests in phase 2 and the performance in the cadet pilot training programme,

whereas, the test scores in phase 1 indicated a moderately positive correlation with the

performance in the cadet pilot training programme. However, the restriction of range as

well as the effect sample size may have contributed to these correlations. Furthermore,

the results of the empirical study supported the acceptance of the null hypotheses.

In the final chapter, conclusions will be drawn on the basis of the results. The limitations

of the research will also be discussed and recommendations made for possible future

research.

208

CHAPTER 6

CONCLUSIONS, LIMITATIONS AND RECOMMENDATIONS FOR FUTURE

RESEARCH

6.1 INTRODUCTION

The general aim was to ascertain the predictive validity of the psychological test battery

used in the selection of cadet pilots for successful completion of the cadet training

programme.

The corresponding general aim and specific aims of the literature review and the

empirical study, as stated in section 1.3 in chapter 1, will guide the discussion in the final

chapter.

6.1.1 General conclusions

In terms of the empirical study, the general aim of this research was to ascertain whether

the psychological tests, which are used for the selection of candidates for the cadet pilot

training programme in a commercial airline using the candidates’ final flying school

results as a criterion measure, are valid.

Based on the findings in the empirical study, it can be concluded that some of the

psychological tests that are used are valid for the prediction of the successful completion

of the cadet pilot training programme in both phases 1 and 2 of the selection process.

Although there was a positive correlation, it was moderately low for most modules of the

flying school results. De Kock and Schlechter (2009, p. 6) concluded in their study that

“since we have shown that sometimes special intelligence (Sn) has incremental validity

over ’g’, the implication is that, practically, such an (small) increase in predictive validity

translates into significant utility yields when considering the low selection ratios, high

costs associated with training and low base rates typical in pilot selection” (Murphy &

Davidshofer in De Kock & Schlechter, 2009). In terms of the current research, this may

imply that testing for specific types of ability and/or intelligence has benefits. However,

the costs associated with training outweigh the benefits.

209

6.1.2 Specific conclusions relating to the literature review

In terms of the literature review, the specific aims were as follows:

(1) To discuss the literature review of the concept of intelligence and aptitude

and explore the nature of these concepts, as well as how intelligence can

be measured using psychological instruments; also to discuss the

concept of selection

(2) To discuss the literature review of the concepts of literacy and numeracy

and describe the dimensions of these concepts; also to discuss the

psychological instruments used to measure literacy and numeracy.

6.1.2.1 Conclusions relating to intelligence, aptitude, psychological tests and selection

The review of literature reported in chapter 2 indicates that intelligence may be explained

as a general factor of intelligence, “g” (Spearman), or in terms of primary abilities, as

described by Thurstone (1938, 1948), or crystallised and fluid intelligence, as proposed

by Cattell. Theories of general factor “g” and separate abilities are combined into a

hierarchical model of intelligence as described by Gustafsson (1984). Carroll’s three-

stratum theory is similar to Cattell’s theory of fluid and crystallised intelligence, except

that it posits a “g” factor at the third stratum.

As a result, a configuration of the battery of tests aligned to the hierarchical model of

intelligence (Gustafsson, 1984) would be appropriate.

In conclusion, it was found that intelligence, whether as a general factor, “g”, or separate

abilities, can be measured using psychological tests, and as a construct, it is significant

in predicting success in the cadet pilot training programme. A similar conclusion was

drawn by Ree and Carretta (1996) in their investigation on general cognitive factor, “g”,

which accounted for 39% of the common variance, while the general psychomotor factor

accounted for 29% of variance, which was consistent with past research by Hunter

210

(1986), Jensen (1993) and Spearman (1904), showing that “g” can be measured in

numerous ways.

De Kock and Schlechter (2009) drew a similar conclusion in their research, which

confirms the widely held belief that measures of general cognitive ability remain

stalwarts in any selection programme for military pilots. They further concluded that their

research has shown that spatial ability can significantly enhance the ability to predict

pilot flight training success. These authors suggest that measures of specific forms of

intelligence (sn), such as spatial ability, could have an incremental validity over “g” and

should therefore not be discarded in favour of measures of general cognitive ability or

fluid intelligence.

Ree and Earles (1991) concluded that the psychometric “g” was the best predictor, and

measures of specific ability were not required to predict training success.

In general, the results of the current study are consistent with previous research on the

prediction of pilot training success in two ways: (1) fluid intelligence remains one of the

best predictors of flight training performance, and (2) the correlations obtained between

predictors and criteria are still only moderate at best (Burke et al., 1997; Damos, 1996;

Hilton & Dolgin, 1991; Hunter & Burke, 1994; De Kock & Schlechter, 2009).

6.1.2.2 Conclusions relating to literacy and numeracy

The review of literature reported in chapter 3, provides support that it is imperative to

examine literacy and numeracy relative to their social function. In South Africa, literacy is

complicated by the fact that knowledge of a second language, usually English, is as vital

for survival and development as the ability to read and write in an African language

(Hutton, 1992). The term “literacy” is therefore used to include basic competency in

English.

Hunter and Burke (1994) asserted that even with better construct data and improved

analytical techniques, differences in cultural context, say, English speaking versus non-

English speaking, may act to moderate validities for specific personality instruments.

211

It is thus concluded that literacy and numeracy skills are vital in the effective

functioning/performance of tasks, including performance in psychological tests and in

any training programme. This conclusion is supported by Downey, Suzuki and Van

Moere (2010, p. 29), who reported that “to prevent radiotelephony miscommunications

in international aviation, ICAO mandated that organizations that employ pilots and air-

traffic controllers have a plan in place to ensure that all such personnel can at least

demonstrate ’operational’ spoken-English proficiency as described in level 4 of the ICAO

rating scale”.

It is concluded that psychological tests that measure intelligence, literacy and numeracy,

are the best for predicting successful completion of the cadet training programme.

6.1.3 Specific conclusions relating to the empirical research

Conclusions, which correspond to the aims of the empirical research, as stated in

section 1.3.3 in chapter 1, will be discussed below regarding the predictive validity of the

psychological test battery used to predict successful completion of the cadet training

programme.

6.1.3.1 Conclusions relating to the psychological test battery

Based on the results of the empirical research reported in chapter 5, it can be concluded

that some of the psychological tests used in the selection battery for pilot selection are

valid for predicting the successful completion of the cadet training programme.

The results of a study by Van der Merwe (2002) indicated that psychometric tests are

not used in isolation, but as an additional aid in decision making and form part of a

defined procedure which includes different, interrelated, specific steps, as well as other

tools.

However, Bartram (2004) postulated that given the ongoing increase in the costs

associated with military and commercial flying, it is not surprising that the search for

more accurate predictions of success in training should continue to receive high priority.

Aptitude test batteries tend to have predictive validities of about .20 to .30 (Hunter &

212

Burke, 1993). Bartram (2004) further stated that if any substantial improvements are to

be made to these levels of validity, it will be necessary to look in domains other than

ability.

6.1.3.2 Conclusions relating to the importance of other variables: English Matric symbol

and literacy, as well as ABET levels, in determining the successful completion of

the cadet training programme

The results of the empirical research reported in chapter 5 show that there is a high

positive correlation with some of the flying school’s modules. In Van der Merwe’s (2003)

research on the extent to which Matric results could be used in a selection battery to

predict future academic performance, there was statistically significant correlation

between the Matric symbol point total (MSPT) and average first-year academic

performance (MPCM 1). The magnitude of the correlation, in terms of effect size, was

between small and medium.

This research supports the conclusion that Matric results have a limited predictive

validity for academic performance. Based on the results of their study, Campbell et al.

(2010) suggested that combining traditional academic ability, emerging psychomotor

assessments and specific personality traits might provide a more empirically derived

account of the “right stuff”.

6.1.4 Potential resolution of the research problem

The conclusion drawn in the current study is that the psychological tests described in the

empirical research (chapter 4) will enhance the selection process for the commercial

airline applicants for the cadet pilot training programme because the tests were found to

be valid for the criteria they are supposed to predict. However, a number of implications

are generated by the present empirical research. While the tests were evaluated as

valid, the manner in which they are used could have an adverse impact. The Uniform

Guidelines of 1978 in the USAS define adverse impact as a situation where there is a

selection rate less than four-fifths (or 80%) of the rate of the group with the highest

selection rate (Arvery & Faley, 1988). Given the differences between the black and white

mean scores in the present study, where the blacks scored lower (table 5.1), there is a

213

possibility that, even though there is no evidence of differential predictions, if tests are

used in a race-neutral, strictly top-down manner, this could have an adverse impact for

black group (Theron, 2009). Theron (2009) suggested that the adverse impact could be

alleviated (but not eliminated) by increasing the predictive validity of the selection

procedure and of the selection ratio.

Given the wide acknowledgement of the need for affirmative action in the developing

potential in a changing South Africa as well as the Employment Equity Act 56 of 1998,

there is increasing pressure on employers to place greater focus on integration across

positions in industry. To this end, South African employers need to develop alternative

selection procedures that will minimise the adverse impact of literacy and numeracy.

At this juncture it is important to re-emphasise that the scope of the present study was

limited and as such the relationships were only evaluated in terms of the predictor and

the criteria, which as indicated earlier, comprised the weighted measures of English

literacy and numeracy and successful completion of the cadet pilot training programme,

as measured by the psychological test battery. As noted in chapter 1, the procedure

adopted in the present study was consistent with the need to facilitate the business

objective of “aiding selection decision making” (Cascio, 1991).

6.2 LIMITATIONS OF THE PRESENT RESEARCH

In the research design, the purpose of the meta-theoretical perspective is to delimit the

boundaries of the research. While this delimitation serves the purpose of focusing and

protecting the researcher, it is restrictive in terms of finding solutions to problems. The

cognitive-behaviourist approach adopted in this research excludes factors such as

motivation, interest, determination and personality characteristics, which may be

hypothesised to be predictive performance in the psychological test battery.

This view is supported by Damos (1996) who stated that no military studies predicting

operational performance were found, although some studies have examined the

concurrent validity of selection tests and operational performance (e.g. Griffin, Morrison,

Amerson & Hamilton, 1987; Griffin & Shull, 1990).

214

6.2.1 Practical constraints

The economic pressures, industrial setting and time constraints tended to influence the

choices which were more practical in the context within which the research was

conducted, as against those that were scientifically and methodologically sound.

However, these limitations had an impact on the design of the research, the aim of which

was to maximise the reliability and validity of the research results.

6.2.2 Restriction of range

Another limitation to be considered is the occurrence of range restriction relating to both

the predictor and criterion variables. According to Cascio (1991), restriction can occur in

the predictor when, for example, only applicants who have survived an initial screening

are considered or when measures are used for selection prior to validation.

This view is shared by Martinussen (1996, p. 15) who reported that “restriction of range

is a major problem in estimating the true validity of psychological measures in pilot

selection, because the selection ratio is very low”. The author stated that restriction of

range can take two forms. Firstly, it can be direct. This occurs when the test in question

is actually used as a selection instrument and only applicants with scores above the cut-

off point are available for further studies. The second form is indirect when the selection

is based upon other predictors that appear to correlate with the test in question.

There may have been range restriction in the present study, because only candidates

who were found suitable in the final stages of the selection process, irrespective of the

test scores, may have been selected for the training programme. Hence this limitation

may have had the effect of lowering the validity estimates obtained for the predictors in

respect of overall effectiveness (Wheeler, 1993).

Cascio (1991) acknowledged that the concurrent design typically ignores the effect of

range restriction. Furthermore, in these designs, the influence of the uncontrolled

variables is also frequently overlooked. In particula, differences caused by motivation

and job experience may have the effect of enhancing or inhibiting the validity of both the

predictor and criteria used, with no way of knowing in advance the direction of such

215

influences (Cascio, 1991). In the current research, interest, motivation and flying

experience may have had either a positive or a negative influence on the scores

obtained by candidates.

According to Ree and Carretta (1996), when validation studies are conducted on

samples that have been systematically selected from the population, there are

restrictions in the variability of predictors (referred to as restriction of range). The above

authors stated that in the study of abilities, the systematic elimination of subjects may

occur as a result of selection on the basis of aptitude tests or personality scores or by

the requirement of having a bachelor’s degree. Restriction of range causes the estimate

of the correlation to be biased downwards.

This view seems to justify the results of this research where the correlations were small.

Ree and Carretta (1996) reported on a study by Thorndike (1949) in which a large group

of applicants were admitted to Army Air Force pilot training without consideration of the

score for the aptitude battery. Correlations with the criterion of pass-fail were computed

for the entire group of applicants and for a subsample that would have passed the

battery of tests on the basis of the selection standards in effect at the end of World War

II. The average decrease in the validity coefficient reported by Thorndike was .29, and

the rank order of validities of tests changed dramatically.

De Kock and Schlechter (2009) reported a similar limitation in their study, in that the

absence of psychometric data on non-successful applicants made estimates of the

population statistics impossible, which is a requirement for computation of adjustments

to the validity coefficient for restriction of range and reliability in the variables.

6.2.3 Sample

The manner in which the sample was selected was also one of the limitations of the

current study.

Ree and Earles (1991), in their study on predicting training success, concluded that in

pooling jobs in validity studies in order to compensate for small sample sizes and using

common regression equations will not cause a drastic reduction in prediction.

216

Carretta and Ree (1996) reported that other studies on information processing found that

it (information processing) could predict pilot training performance, whereas their study

using the BAT tests based on information processing was surprising. They concluded

that the lack of validity may have been the result of sampling error, but the fact that the

BAT tests were not incrementally valid, was not surprising because of their degree of “g”

saturation.

Flotman (2002) noted in his study that only candidates who had passed the

psychometric evaluation (the cut-off point determined by the Military Psychological

Institute, based on specific norms for black and white candidates) were included in the

Ground School Phase training.

The sample in the current study was preselected by the recruitment centre, based on the

information provided either on the application form or in the letter of application. The

actual criteria according to which the recruitment staff would have nominated the

individual applicants cannot be determined and may have ranged from personal

knowledge or the pre-screening interview to various other interpersonal constructs and

other objective or subjective criteria.

6.3 RECOMMENDATIONS

It is recommended that the following issues be incorporated into further research on the

prediction of successful completion of a training programme using psychological test

battery:

6.3.1 Incorporation of other psychological paradigms

When conducting research on predicting the successful completion of a training

programme using a psychological test battery, it would be beneficial to incorporate

variables from other paradigmatic perspectives such as job experience, personality,

attitudes and motivation.

217

This view is supported by Martinussen (1996), who recommended that a selection

battery for pilots should include tests measuring cognitive and psychomotor/information

processing abilities, as well as knowledge of aviation. Martinussen (1996) further

suggested that there should be less emphasis on tests of general intelligence and

academics and since the utility of personality tests is uncertain, they should not be

automatically be used in pilot selection.

Carretta et al. (1996, p. 31), in their study on the prediction of situation awareness in F15

pilots, “found that the first-order factors measuring the constructs verbal working

memory, spatial working memory, spatial reasoning, divided attention, aiming, reaction

time, and rate control showed validity”. They therefore recommended that test

measuring these constructs should be investigated for inclusion in an enhanced pilot

selection battery. It can be inferred from this statement that there are other constructs

that could enhance pilot selection in both the military and commercial environments.

Carretta and Ree (1996) recommended that future measurements of pilot aptitude,

which have been shown to be highly “g” loaded, could include tests based on cognitive

components, chronometric methods or neural conductive velocity, because the

incremental validity of “g” specific cognitive abilities (e.g. verbal, qualitative, spatial or

perceptual speed) has been shown to be small or nonexistent. However, the

incremental validity of pilot job knowledge such as knowledge of aircraft concepts,

instruments, principles and terms, psychomotor abilities and personality scores has also

been found to be small but significant.

Taking into account the influence of measures of literacy and numeracy results found in

the present study, it is recommended that such variables be included to enhance the

predictive power of such tests. Although the results of the present study indicate that the

literacy and numeracy scores obtained provide an indication of proficiency, from a

language perspective, this does not necessarily mean that the paricipants will perform

better in the psychological test battery and the training programme. Furthermore, the

incorporation of tests, such as the Learning Potential Computerised Adaptive Test

(LPCAT) could enhance the selection process. In a study conducted by De Beer (2006)

on dynamic testing, the results of this study showed that the LPCAT provided useful

information by indicating the level of general reasoning ability and learning potential of

218

an individual. It was also reported that LPCAT may indicate the academic level an

individual is likely to perform at or the amount of effort required from an individual to

achieve success at a particular level.

This view was emphasised by Carretta (1996) in the statement that new cognitive

components offer measurement of general cognitive ability with almost no content in the

usual sense because they do not require previous learning other than the language

requirements of the instruction. In the author’s view the problem of an adverse impact on

minorities and women could be reduced or avoided if the new cognitive tests were to be

added to the pilot selection system.

6.3.2 Cross-validation

It is recommended that cross-validation of the psychological tests be undertaken on a

second sample of the job applicants for the cadet pilot training programme. This is

essential, but rarely happens (Murphy, in Cascio, 1991). Cross-validations are preferred

for the following reasons:

All of the information contained in a sample can be used. This maximises the stability

of regression weights.

Statistical correction is faster and requires less effort on the part of the researcher.

Formula estimates appear to be highly accurate.

Coupled with this is the need to select the sample in such a way that it is possible to

minimise, even eliminate, pre-selection and restriction of range.

6.3.3 Longitudinal study of progress through performance

Owing to the existence of substantial group differences in average test scores,

particularly differences between black and white job applicants, the issue of fairness

needs to be addressed. A test may be fair in predicting performance, but still predict

performance rather poorly.

219

Human performance is far too complex to expect anything approaching perfect

prediction. Tests are at best only moderately good predictors of job performance

(Hartigan & Wigdor, 1989). One of the consequences of prediction errors is that certain

people who could perform well on the job, but who score in the lower ranges in the tests,

are screened out, whereas others who score well on the tests, and are thus selected,

could perform inadequately on the job (Hartigan & Wigdor, 1989).

It is further recommended that the effect of literacy levels, particularly English literacy

and numeracy, in psychological tests should be explored. Also, a thorough job analysis

should be conducted to identify critical skills and competences for a pilot, thus ensuring

that appropriate and relevant psychological instruments are used in the selection

process.

De Kock and Schlechter (2009) appeared to share similar sentiments when they

asserted that selection decision errors in pilot selection may be catastrophic. Such errors

could be minimised by using more accurate selection procedures that capture more of

the factors that influence performance in the cockpit.

With the advances in research in the military environment, it is strongly recommended

that future research on pilot selection be conducted in collaboration with the SAAF,

acknowledging that there could be differences in some of the specific variables these

environments would require of their personnel.

6.4 CHAPTER SUMMARY

The conclusions drawn on the basis the specific aims of the research were presented in

this chapter. The conclusions guided by the corresponding general and specific aims of

the literature review and the empirical research were also discussed.

The limitations of the present study were discussed in terms of practical constraints,

restriction of range and the sample, in determining the predictive validity of the

psychological tests, in so far as they hinder efforts to maximise the validity of the present

study. The results of a single study of this kind do not necessarily provide conclusive

evidence of the reliability and prediction power of the psychological tests used. However,

the findings of the present research study do form a foundation for further research.

220

Recommendations for further research were presented, taking into account the

incorporation into the research of other psychological perspectives and paradigms from

other disciplines. Cross-validation (and the reason for it) was recommended. The effect

of the longitudinal studies through literacy training programmes and performance, as

well as the contribution of literacy, numeracy and aptitude at a micro-economic level (i.e.

in terms of productivity), should be also be considered.

221

REFERENCE LIST

Aiken, L. R. (1991). Psychological testing and assessment (7th ed.). Boston: Allyn and

Bacon.

Aiken, L. R. (1997). Psychological testing and assessment (9th ed.). Boston: Allyn and

Bacon.

Aiken, L. R. (2003). Psychological testing and assessment (11th ed.). Boston: Pearson

Education Group, Inc.

Airline pilot job description and activities. (May 17, 2010). Retrieved from

http://www.prospects.ac.uk/p/types_of_job-description.jsp

Anastasi, A. (1990). Psychological testing (6th ed.). New York: MacMillan.

Anastasi, A. & Urbina, S. (1997). Psychological testing (7th ed.). New Jersey: Upper Saddler

River, Prentice Hall.

Anderson, J. (1976). Psycholinguistic experiments in foreign language testing. St Lucia,

Queensland: University of Queenslands Press.

Armour-Thomas, E. & Gopaul-McNicol, S. (1998). Assessing intelligence: Applying a bio-

cultural model . Thousand Oaks: Sage Publications.

Arvey, R.D. & Faley, R.H. (1988). Fairness in selecting employees (2nd ed.). Reading,

Mass.: Addison-Wesley.

Aspeling, E. G. (1990). Vlieenierskeuringstrategiee vir die 1990s. Pretoria: National Institute

for Personnel Research.

222

Avenant, T. J. (1988). The establishment of an individual intelligence scale for adult South

African. Report on an exploratory study conducted with the WAIS-R on a sample of

black prison warders (Report on an exploratory study conducted with the WAIS-R on a

sample of black prison warders No. P-91). Pretoria: Human Sciences Research Council.

Barkhuzen, W., Schepers, J., & Coetzee, J. (2002). Rate of information processing and

reaction time of aircraft pilots and non-pilots. South African Journal of Industrial

Psychology, 28(2), 67-76.

Bartholomew, D. J. (2004). Measuring intelligence. Cambridge: Cambridge University Press.

Bartram, D. (2004). Assessment in organizations. Applied Psychology: An International

Review, 53(2), 237-259.

Bartram, D. & Baxter, P. (1996). Validation of the Cathay pacific airways pilot selection

program. International Journal of Aviation Psychology, 6(2), 149-169.

Bedell, B., Van Eeden, R. & Van Staden, F. (1999). Culture as a moderator in psychological

test performance: Issues and trends in South Africa. South African Journal of Industrial

Psychology, 25(3), 1-7.

Bergh, Z. (2003). Methodology. In Z.C. Bergh & A.L. Theron, (Eds). Psychology in the work

context. (2nd ed., pp 17-43). Cape Town: Oxford University press, Southern Africa

(Pty) Ltd.

Bergh, Z. (2009). Assessment of personality and individual differences. In Z.C. Bergh, &

A.L. Theron (Eds.), Psychology in the work context (4th ed., pp. 406-432). Cape Town:

Oxford University press, Southern Africa (Pty) Ltd.

223

Bingham, W. (1937). Aptitude and aptitude testing. New York, N.Y.: Harper.

Birch, A. & Hayward, S. (1994). Individual differences. Houndmills: MacMillan.

Bloom, L. & Lahey, M. (1978). Language development and language disorder. New York:

MacMillan.

Blox Test Administrator manual. (1983). The National Institute of Personnel Research.

Johannesburg: Human Sciences Research Council.

Bohlmann, C.A. & Pretorius, E.J. (2002). Reading skills and mathematics. South African

Journal of Higher Education, 6(3), 196-206.

Bornstein, R., Schuster, C. S. & Ashburgn, (1992). Perception without awareness: cognitive,

clinical and social perspective. New York: Guilford Press.

Brace, N., Kemp, R. & Snelgar, R. (2009). SPSS for psychologists (4th ed.). Basingstole:

Palgrave Macmilla.

Breakwell, G. M., Hammond, S. & Fife-Schaw, C. (2000). Research methods in psychology.

London ; Thousand Oaks, Calif.: Sage Publications.

Burke, E., Hobson, C., & Linsky, C. (1997). Large sample validations of three general

predictors of pilot training success. International Journal of Aviation Psychology, 7(3),

225-234.

Butcher, J.N. (2002). Assessing pilots with ‘the wrong stuff’: A call for research on

emotional health factors in commercial aviators. International Journal of Selection and

Assessment, 10(1/2), 168- 184.

224

Calfee, R. (1993). Paper, pencil, potential and performance. Current Directions in

Psychological Science, 2, 6-7.

Campbell, J. S., Castaneda, M., & Pulos, S. (2010). Meta-analysis of personality

assessments as predictors of military aviation training success. The International

Journal of Aviation Psychology, 20(1), 92-109.

Careers offered by SAA (2010, October 21). Retrieved from http://www.caa.co.za./careers

in aviation.

Carpenter, P. A., Just, M. A. & Shell, P. (1990). What one intelligence test measures: A

theoretical account of the processing in the raven progressive matrices

test. Psychological Review, 97, 404-431.

Carretta, R.T. & Ree, M.J. (2003). Pilot selection methods. In P.S. Tsang & M.A. Vidulich

(Eds.), Principles and practice of aviation psychology (pp. 357-396). Mahwah:

Lawrence Erlbaum Associates, Inc.

Carretta, T. R. & Ree, M. J. (1996). U.S. Air force pilot selection tests: What is measured

and what is predictive? International Journal of Aviation Psychology, 67(3), 279-282.

Carretta, T. R., Perry, D.C. Jr. & Ree, M. J. (1996). Prediction of situation awareness in F-15

pilots. International Journal of Aviation Psychology, 6(1), 21-41.

Carretta, T.M. (1992). Recent developments in U.S, Air Force pilot candidate selection and

classification. Aviation, Space and Environmental Medicine, (December), 1112-1114.

Carretta, T.R. & Ree, M.J. (1989). Pilot - candidate selection method: Sources of

validity. International Journal of Aviation Psychology, 4(2), 103-117.

225

Carretta, T.R. & Ree, M.J. (1994). Pilot-candidate selection methods: Sources of

validity. The International Journal of Aviation Psychology, 4(2), 103-117.

Carretta, T.R. & Ree, M.J. (1996). U.S. air force pilot selection tests: What is measured and

what is predictive? Aviation, Space and Environmental Medicine, 67(3), 279-283.

Carroll, J. B. (1993). Human cognitive abilities: A survey of factor analytic studies.

Cambridge, England: Cambridge university press.

Cascio, W. F. (1991). Applied psychology in personnel management (4th ed.). Englewood

Cliff, New Jersey: Prentice-Hall.

Cashdan, A. (1986). Literacy: Teaching and learning language skills. Oxford, UK New York,

NY, USA : Blackwell.

Cattell, R. B. (1987). Intelligence: Its structure, growth and action. Amsterdam New York :

North-Holland New York: Elsevier Science Pub. Co.

Chew, C. R. (1992). Policies for literacy. In C. Hedley, D. Feldman & P. Antonacci

(Eds.), Literacy across the curriculum (pp. 3-14). Norwood, New Jersey: Ablex

Publishers. Corp.

Clark, H. H. (1996). Using language. England; New York: Cambridge University press.

Clayton, K. N. (1984). An introduction to statistics for psychology and education. Columbus:

C.E. Merrill Pub. Co.

Cohen, B. H. (2008). Explaining psychological statistics (3rd ed.). New Jersey: John Wiley

and Sons, Inc.

226

Cohen, J. (1992). Qualitative methods in psychology: A power primer. Psychological

Bulletin, 112(1), 155-159.

Cohen, R.J. & Swerdlink, M.E. (2002). Psychological testing and assessment: An

introduction to tests and measurement (5th ed.). Mountain View, Calif.: McGraw-Hall

Companies, Inc.

Cooper, D.R. & Schindler, P.S. (1998). Business research methods (6th ed.). Boston:

Irwin/McGraw-Hill.

Copans, G. (2007). Clear skies for South African aviation industry. (2010, April 31)

Retrieved from http://www.engineeringnews.co.za/article/clear-skies-for - South-

African-aviation industry.

Copeland, R. W. (1979). How children learn mathematics: Teaching implications of Piaget’s

research (3rd edition ed.). New York: MacMillan.

Cramer, R. L. (1978). Writing, reading and language growth: An introduction to language

arts. Columbus: Merill.

Cronbach, L. J. (1990). Essentials of psychological testing (5th ed.). New York: Harper

Collins Publishers, Inc.

Cunningham, G. K. (1986). Educational and psychological measurement. New York: Colliers

MacMillan.

Curwin, J. & Slater, R. with Hart, M. (1994). Numeracy skills for business. London, Chapman

Hall New York: Wisley.

227

Damos, D. L. (1996). Pilot selection batteries: Shortcomings and perspectives. The

International Journal of Aviation Psychology, 6(2), 199-209.

De Beer, M. (1994) Dynamic assessment – meeting new challenges in training and

development. Paper presented at the Sixth Conference of the South African Institute

for Management Scientists, University of the Western Cape, Cape Town.

De Beer, M. (1997). The construction of a dynamic, computerized adaptive test for

measuring Learning Potential (Preliminary report) Pretoria: Human Sciences Research

Council.

De Beer, M. (2000). The construct and evaluation of a dynamic computerised adaptive tests

for the measurement of learning potential. (Unpublished Doctoral thesis). University of

South Africa, Pretoria.

De Beer, M. (2006). Dynamic testing: Practical solutions to some concerns. South African

Journal of Industrial Psychology, 32(4), 8-14.

De Kock, F. & Schlechter, A. (2009). Fluid intelligence and spatial reasoning as predictors of

pilot training performance in the South African air force (SAAF). South African Journal

of Industrial Psychology, 35(1), 1-8.

De Montalk, R. J. (2008). Developing proficiency in air transport pilots: The case for

introduction of non - technical skills in basic pilot training programmes. (Unpublished

Doctoral thesis). Massey University Palmerston North, New Zealand.

De Vito, J. A. (1973). Language: Concepts and processes. Englewood Cliffs, New Jersey:

Prentice-Hall.

228

Department of Education. (1997). Numeracy learning programmes document: Foundation

phase. Pretoria: Department of Education.

Dillon, R. F. (1997). Handbook on testing. Westport, Conn.: Greenwood Press.

Downey, R., Suzuki, M. & Van Moere, A. (2010). High-stakes English-language assessments

for aviation professionals: Supporting the use of a fully automated test of spoken-

language proficiency. IEEE Transactions on Professional Communication, 53(1), 18-32.

Elkonin, D., Foxcroft, C., Roodt, G., & Astbury, G. (2005). The use of assessment measures

in various applied contexts. In C. Foxcroft, & G. Roodt (Eds.), An introduction to

psychological assessment in South African context (2nd ed., pp. 195-212). Cape Town:

Oxford University press Southern Africa.

Ellsworth, N.J., Hedley C.N. & Baratta N.A. (Eds). (1994). Literacy: a redefinition. Hillsdale:

NJ. L. Erlbaum Associate Publishers.

Employment Equity Act, 55, (1998). Government Gazette (19370).

Equal Employment Opportunity Commission (EEOC). (1978). Uniform Guidelines on

Employee Selection Procedures. Federal Register, 12333-12336.

Everette, J. E. (1991). Tertiary entrance predictors of first year university

performance. Austrian Journal of Education, 35(1), 24-40.

Feldhusen, J.F. & Jarwan, F. (1995). Predictors of academic success at state-supported

residenial school for mathematics and science: A validity study. Educational and

Psychological Measurement, 55, 505-512.

229

Ferdman, B. M., Weber, R. & Ramirez, E.D. (1994). Literacy across language and culture.

Albany, New York: State University of New York press.

Field, A. (2000). Discovering statistics using SPSS for windows: Advanced techniques for

beginners. London: Sage Publications Limited.

Field, A. P. (2005). Discovering statistics using SPSS: (and sex, drugs and rock 'n' roll).

London: Sage Publications.

Fillmore, C. J., Kempler, D. & Wang, W. (1979). Individual differences in language ability

and language behavior. New York: Academic press.

Flotman, A. P. (2002). The predictive validity of the selection battery for trainee pilot in the

South African air force. (Unpublished Master’s dissertation). University of South Africa,

Pretoria.

Foxcroft, C. & Roodt, G. (2005). An overview of assessment: Definition and scope. In C.

Foxcroft & G. Roodt (Eds.), An introduction to psychological assessment in the South

African context (2nd ed., pp. 3-7). Cape Town: Oxford university press, South Africa

(Pty) Ltd.

Foxcroft, C., Roodt, G. & Abrahams, F. (2005). Psychological assessment: A brief

retrospective overview. In C. Foxcroft & G. Roodt (Eds.), An introduction to

psychological assessment in the South African context (2nd ed., pp. 8-23). Cape Town:

Oxford university press, South Africa (Pty) Ltd.

230

Foxcroft, C.D. & Aston, S. (2006). Critically examining language bias in the South African

adaption of the WAIS-III. South African Journal of Industrial Psychology, 32(4), 97-

102.

Freedman, G. (2000). Validity of measures of cognitive ability for predicting literacy and

numeracy performance (Unpublished Master’s dissertation). Pretoria: University of

South Africa.

Ghiselli, R.D., Campbell, J.P. & Zedeck, S. (1981). Measurement theory for the behavioural

sciences. San Francisco, CA: Freeman.

Glass, G.V. & Stanley, J.C. (1984). Statistical methods in education and psychology

(2nd Ed.). Englewood Cliffs, N.J.: Prentice-Hall.

Green, B. (1999). The new literacy challenges? literacy learning: Second thoughts. A

Journal of the Australian Literacy Educator's Association, 7(1), 36-46.

Griffin, G., Morrison, T., Amerson, T. & Hamilton, P. (1987). Predicting air combat

maneuvering (ACM) performance: Fleet fighter ACM readiness program grades as

performance criteria (Tec. Rep. No. NAMRL-1333). Pensacola, FL: Naval Aerospace

Medical Research Laboratory.

Griffin, G. & Shull, R. (1990). Predicting F/A-18 fleet replacement squadron performance

using an automated battery of performance-based tests (Tech. Rep. No. NAMRL-

1354). Pensacola, FL: Naval Aerospace Medical Research Laboratory.

Groth-Marnat, G. (2003). Handbook of psychological assessment (4th ed.). New Jersey:

John Wiley and Sons, Inc.

231

Gustafsson, J. E. (1989). A unifying model for the structure of intellectual

abilities. Intelligence, 8, 179-203.

Halliday, M. A. K. (1973). Explorations in the functions of language. London: Edward Arnold.

Halliday, M. A. K. (1976). System and function in language: Selected papers. London:

Oxford University Press.

Halliday, M. A. K. & Mathiessen, C. M. I. M. (1999). Construing experience through

meaning: A language-based approach to cognition. London: Cassell.

Hamon, P. (2000). Reflecting on literacy in education. London; New York: Falmer Press.

Harris, T.L. & Hodges, R.E. (Eds.). (1995). The literacy dictionary. Newark, Del.:

International Reading Association ERIC Clearinghouse on Reading and Communication

Skills, National Institute of Education.

Hartigan, J.A. & Wigdor, A.K. (Eds.). (1989). Fairness in employment testing: Validity

generation, minority issues, and the general aptitude test battery. Washington D.C.:

National Academy press.

Hasan, R. & Martin J.R. (1989). Language development: Learning language, learning

culture, meaning and choice in language. Norwood, N.J.: Ablex Pub. Corp.

Hesketh, B. & Robertson, I. (1993). Validating personnel selection: A process model for

research and practice. International Journal of Selection and Assessment, 1(1), 3-17.

Heugh, K. (2001). Many languages in education. Perspectives in Education, 19(2), 17.

232

Hilton, T. F. & Dolgin, D. L. (1991). Pilot selection in the military of free world. In R. Gal, &

A. D. Mengelsdorf (Eds.), Handbook of military psychology (pp. 81-101). New York:

Wiley.

Holburn, P. T. (1992). Blox test administrator's manual. Pretoria: Human Sciences Research

Council.

Holburn, P. T. (1993). Psychometric assessment: Identifying appropriate assessment

techniques for specific jobs to ensure that the right person is selected. Paper presented

at the Congress in culture fair assessment.

Holme, R. (2004). Literacy: An introduction. Edinburgh: Edinburg University Press Ltd.

Hoole, C. & Vermuelen, L.P. (2003). Job satisfaction among South African aircraft

pilots. South Journal of Industrial Psychology, 29(1), 52-57.

Hough, B. & Horne, T.J. (2006a) ELSA-FET Intermediate Assessment Administration.

Unpublished.

Hough, B. & Horne, T.J. (1989). English literacy skills assessment (test manual). Natal:

Pinetown Printers.

Howitt, D. & Cramel, D. (2003). An introduction to statistics in psychology (Rev. 2nd ed.).

England: Pearson Education Limited.

Hughes, M. (1986). Children and numbers: Difficulties in learning mathematics. Oxford

[Oxfordshire], UK New York, NY, USA: B. Blackwell.

Hunter, D.R. & Burke, E.F. (1994). Predicting aircraft pilot-training success: A meta-analysis

of publish research. The International Journal of Aviation Psychology, 4(4), 297-313.

233

Hunter, J. E. (1986). Cognitive ability, cognitive attitude, for knowledge and job

performance. Journal of Vocational Behavior, 29, 340-362.

Hutchison, K. (March, 2009). Airline pilot: Entry requirements. Retrieved August, 07, 2010,

from http//www.prospects.ac.uk/p/types_of_job/airline_entry_requirements.jsp

Hutton, B. (Ed.). (1992). Adult basic education in south africa: Literacy, english as a second

language and numeracy. London: Oxford University press.

Huysamen, G. K. (1996). Psychological measurement: An introduction with South African

examples (3rd Rev. Ed.). Pretoria: van Schaik.

Insua, A. M. (1983). WAIS-R factor structures in two cultures. Journal of Cross-Cultural

Psychology, 14(4), 427-438.

Jackson, C. (1996). Understanding psychological testing. Leicester: The British Psychological

Society.

Jensen, A. R. (1993b). Test validity: G versus "tactical knowledge". Current Directions in

Psychological Science, 2, 9-10.

Johnston, N. (1996). Psychological testing and pilot licensing. The International Journal of

Aviation Psychology, 6(2), 179-197.

Jukes, M.C.H. & Grigorenko, E.L. (2010). Assessment of cognitive abilities in multiethnic

countries: The case of the Wolof and Mandinka in the Gambia. British Journal of

Educational Psychology, 80, 77-97.

Kerlinger, F. N. (1986). Foundations of behavioural research (3rd ed.). New York: CBS.

234

Kerlinger, F.N. & Lee, H.B. (2000). Foundations of behavioural research (4th ed.). USA:

Harcourt College Publishers.

Kline, P. (1998). The new psychometrics: Science, psychology and measurement. London:

Routledge.

Kriek, H. & Dowdeswell, K. (2010). Adverse impact in South Africa. In J. M. Outtz

(Ed.), Adverse impact: Implications for organizational staffing and high stakes

selection (pp. 375-399). New York: Routledge, Taylor and Francis Group.

Kriek, N. J. S. (1993). In a memo to Mnr. T.J. Horne (Ed.), English literacy skills assessment

(ELSA) project predictive validity at STD 6 level. Pretoria:

KrÜgel, R. (2006). The influence of the language proficiency of English teachers who are not

native speakers of English on language skills of their learners. (Unpublished Master’s

dissertation). North-West University, Vanderbijpark.

Labour Relations Act, No 66 (1995). Government Gazette 366 (16861).

Leong, F.T.L. & Austin, J.T. (2006). The psychology research book: A guide for graduate

students and research assistants. Thousand Oaks, Califonia: Sage Publications.

Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York, Oxford: Oxford

University press, Inc.

Lieberman, P. (1975). On the origins of language: An introduction to the evolution of human

speech. New York: MacMillan.

235

Loukopoulos, D., Dismukes, R.K. & Barshi, I. (2003, April). Concurrent task demands in the

cockpit: Challenges and vulnerabilities in routine flight operations. 12th International

Symposium on Aviation Psychology. 1- 6.

Louw, D. A. & Edwards, D. J. A. (1997). Psychology an introduction for students in South

Africa (2nd ed.). Johannesburg: Heinemann.

Lukin, A. & Ross, L. (1997). The numeracy handbook: A resource for literacy and numeracy

teachers. Sew South Wales: New South Wales AMES: NCELTR.

Marais, H. The impact of global recession on South Africa. (2010, June, 21). Retrieved from

http://www.amandlapublishers.co.za/home-menu-item/156

Martinussen, M. (1996). Psychological measures as predictors of pilot performance: A meta-

analysis. The International Journal of Aviation Psychology, 6(1), 1-20.

Martinussen, M. & Torjussen, T. (1998). Pilot selection in Norwegian air force: A validation

and meta-analysis of the test battery. The International Journal of Aviation

Psychology, 8(1), 33-45.

Mazak, C. (2007). Appropriation and resistance in the (English literacy practices of Puerto-

rican farmers. In V. Purcell-Gates (Ed.), Cultural practices of literacy: Case studies of

language, literacy, social practice and power. (pp. 25-40). Mahwah, N.J.: Lawrence

Erlbaum Associates, Inc.

McClelland, D.C. (1993). Intelligence is not the best predictor of job performance. Current

Directions in Psychological Science, 2, 5-6.

236

McCulloch, S. (1993). Professional forum: Recent trends in international

assessment. Internal Journal of Selection and Assessment, 1(1), 59-61.

Meiring,D., Van de Vijver, A.J.R., Rothman, S. & Barrick, M.R. (2005). Construct, item, and

method bias of cognitive and personality tests in South Africa. South African Journal of

Industrial Psychology, 31 (1), 1-8.

Moerdyk, A. (2009). The principles and practice of psychological assessment (First ed.).

Pretoria: Van Schaik Publishers.

Molosiwa, A. (2007). Language and literacy in Botswana. In V. Purcell-Gates (Ed.), Cultural

practices of literacy: Case studies of language, literacy, social practice and power (pp.

41-54). Mahwah, N.J.: Lawrence Erlbaum Associates, Inc.

Morgan, G. (1980). Paradigm, metaphors and puzzle solving in organizational

theory. Administrative Science Quarterly, 25(4), 605-622.

Moss, B. J. (1994). Literacy across communities. Cresskill, N.J.: Hampton Press.

Mouton, J. & Marais, H.C. (1990). Basic concepts in methodology of the social sciences.

Pretoria: Human Sciences Research Council.

Muller, J. & Schepers, J. (2003). The predictive validity of the selection battery used for

junior leaders training within the South African National Defense Force. South African

Journal of Industrial Psychology, 29(3), 87-98.

Mundy-Castle, C. A. (1974). Social and technological intelligence in western and non-

western cultures. In S. Pilowsky (Ed.), Cultures in collision (pp. 46-52). Adelaide:

Australian National Association for Mental Health.

237

Murphy, K. R. (1989). Is the relationship between cognitive ability and job performance

stable over time? Human Performance, 2, 183-200.

Murphy, K.R. & Davidshofer, C.O. (2005). Psychological testing: Principles and

applications (6th ed.). Upper Saddle River, New Jersey: Pearson/Prentice Hall.

Nell, V. (1999). Standardising the WAIS-III and WMS-III for South Africa: Legislative,

psychometric and policy issues. South African Journal of Psychology, 24(2), 100-108.

Nunan, D. (1999). Second language teaching and learning. Boston, Mass.: Heinle & Heinle

Publishers.

Nunns, C. & Ortlepp, K. (1994). Exploring predictors of academic success in psychology 1 at

Wits University as an important component of fair student selection. South African

Journal of Psychology, 24, 201-207.

Nzama, L., De Beer, M. & Visser, D. (2008). Predicting work performance through selection

interview rating and psychological assessment. South African Journal of Industrial

Psychology, 34(3), 39-47.

O'Hare, D. & Walsh, K. (2003). The shape of aviation psychology II: Inspiration and

continuation. International Journal of Aviation Psychology, 13(1), 23-32.

Oller, J. W. (1979). Language testing at school: A pragmatic approach. Harlow, Essex:

Longman.

Oller, J.W. & Perkins, K. (1980). Research in language testing. Boston: Newbury House.

238

Owen, K. (1991). Test bias: The validity of the junior aptitude tests (JAT) for various

population groups in South Africa regarding construct measured. South African Journal

of Psychology, 21(2), 112-118.

Owen, K. (1992). The suitability of raven's standard progressive matrices for various groups

in South Africa. Personality, Individual Differences Journal, 13(2), 149-159.

Owen, K. (1996). Aptitude tests. In K. Owen, & J. J. Taljaard (Eds.), Handbook for the use

of psychological and scholastic tests of the HRSC (pp. 191-270). Pretoria: Human

Science Research Council.

Owen, K. & Taljaard, J.J. (1989). Handbook for the use of psychological and scholastic test

of the IPER and NIPR. Pretoria: Human Sciences Research Council.

Papalia, C.K. & Olds, S.W. (1995). Human development (6th ed.). New York: McGraw-Hill.

Paterson, H. & Uys, K. (2005). Critical issues in psychological test use in the South African

workplace. South African Journal of Industrial Psychology, 31(2), 12-22.

Pierce, C. A., Block, R.A. & Aguinis, H. (2004). Cautionary note on eta-squared values from

multifactor ANOVA designs. Educational and Psychological Measurement, 64(6), 916-

924.

Pretoruis, E. & Naude, H. (2002). A culture in transition: Poor reading and writing ability

among children in South African township. Early Child Development and Care, 172(5),

439-449.

Price, H. (1997). The value of managing language diversity. People Dynamics, (January),

20-22.

239

Purcell-Gates, V. (Ed.). (2007). Cultural practices of literacy: Case studies of language,

literacy, social practices and power. Mahwah, N.J.: Lawrence Erbaum Associates

Publishers.

Pyle, D. W. (1979). Intelligence: An introduction. Boston: Routledge and Kegan Paul.

Raven, J. (2000). The Raven's progressive matrices: Change and stability over culture and

time. Cognitive Psychology, 41, 1-48.

Raven, J.C., Raven, R.C. & Court, J.H. (1998). Raven’s manual. Aseford: Oxford

Pyschologists press Ltd.

Raven, J.C., Court, J.H. & Raven, J. (1988). Manual for Raven’s Progressive Matrices and

Vocabulary Scales (Section 4). London: H.K. Lewis.

Raven, J.C. & Court, J.H. (1985). Manual for Raven’s progressive matrices. London: H.K.

Lewis.

Ree, M. J. & Carretta, T. M. (1996). Central role of g in military pilot selection. The

International Journal of Aviation Psychology, 6(2), 111-123.

Ree, M. J. & Earles, J. A. (1992). Intelligence is the best predictor of job

performance. Current Directions in Psychological Science, 1, 86-89.

Ree, M. J. & Earles, J. A. (1993). g is to psychology what carbon is to chemistry: A reply to

Sternberg and Wagner, McClelland and Calfee. Current Directions in Psychological

Science, 2, 11-12.

Ree, M.J. & Carretta, T.R. (1996). Central role of g in military pilot selection. International

Journal of Aviation Psychology, 6(2), 111-123.

240

Ree, M.J. & Carretta, T.R. (2002). g2K. Human Performance, 15(1), 3-23.

Ree, M.J. & Earles, J.A. (1991). Predicting training success: Not much more than

g. Personnel Psychology, 44, 321-332.

Reschly, D. J. & Robinson-Zanartu, C. (2002). In G. Goldstein, & M. Hersen

(Eds.), Handbook of psychological assessment (pp. 183-186). New York: Elsevier

Science.

Retzlaff, P. D. & Gibertini, M. (1987). Air force pilot personality: Hard data on the "right

stuff". Multivariate Behavioral Research, 22, 383-399.

Rogers, S. (1975). Children and language: Readings in early language and socialization.

London: Oxford University Press.

Rushton, J. P., Skuy, M. & Fridjhon, P. (2003). Performance on Raven's advanced

progressive matrices by African, East Indian, and white engineering students in South

Africa. Intelligence, 31(2), 123-137.

Samunda, R. J., Feuerstein, R., Kaufman Lewis, A. S. & Stenberg, R. A. (1998). Advances in

cross-cultural assessment. Thousand Oaks, California: Sage Publication.

Sattler, J. M. (1988). Assessment of children (3rd ed.). San Diego: J.M. Sattler.

Schepers, J.M. (1983), South African Wechsler Adult Intelligence Scale. (Manual ed.).

Pretoria: Human Sciences Research Council.

Schmidt, F.L. & Hunter, J.E. (1998). The validity and utility of selection methods in

personnel psychology and theoretical implications of 85 years research

findings. Psychological Bulletin, 124(2), 262-274.

241

Schoeman, H. S. (1991). In Messrs B. Hough & T.J. Horne (Eds.), English literacy skills test

(personal communication- memo, April 26, 1991). Pretoria:

Segal, M. H., Dasen, P. R., Berry, J. W. & Poortinga, Y. H. (1999). Human behavior in global

perspective (2nd ed.). Needham Heights M.A.: Allyn and Bacon.

Shaughnessy, J. J. (2000). Research methods in psychology. Boston: McGraw-Hill.

Shochet, I. M. (1994). The moderator effect of cognitive modifiability on a traditional

undergraduate admissions test for disadvantaged black students in South Africa. South

African Journal of Psychology, 24, 208-215.

Shum, D., O'Gorman, J. & Myors, B. (2006). Psychological testing and assessment. South

Melbourne, Australia; New York: Oxford university press.

Shuttleworth-Edwards, A. B., Kemp, R. D., Rust, A. L., Muirhead, J. G. L., Hartman, N.P. &

Radloff, S.E. (2004). Cross-cultural effect on IQ test performance: A review and

preliminary normative indications on WAIS-!!! test performance. Journal of Clinical and

Experimental Neuropsychology, 26(7), 903-920.

Shuttleworth-Jordan, A. B. (1996). On not reinventing the wheel: A clinical perspective on

cultural relevant test usage in South Africa. South African Journal of Psychology, 26(2),

96-102.

Skills Developments Act, 97, (1998) Government Gazette, (19420).

Society for Industrial Psychology. (1992). Guidelines for the validation and use of personnel

selection procedures. Psychological Association of South Africa.

242

Sommer, M., Olbrich, A. & Arendasy, M. (2004). Improvements in personnel selection with

neural networks: A pilot study in the field of aviation psychology. International Journal

of Aviation Psychology, 14(1), 103-115.

South African Airways. (2010) . South African Airways Employment opportunities. Retrieved

from http://www.flysaa.com.

Southwood, S., Spannerberg, R. & Stoker, J. (1997). Developing number sense. Pretoria:

Via Africa.

Spearman, C. (1904). General intelligence objectively determined and measured. American

Journal of Psychology, 15(2), 201-203.

SPSS base 9.0 application guide (1999). Chicago: SPSS Inc.

Statistics South Africa. (2006). Population statistics: Mid-year population

estimates. Retrieved from http://www.statssa.gov.za.

Stauffer, J. & Ree, M.J. (1996). Predicting with logistic or linear regression: Will it make a

difference in who is selected for pilot training. International Journal of Aviation

Psychology, 6(3), 233-240.

Sternberg, R. J. (1982). Handbook of human intelligence. Cambridge (Cambridgeshire) New

York: Cambridge University press.

Sternberg, R.J. & Wagner, R.K. (1993). The g-ocentric view of intelligence and job

performance is wrong. Current Directions in psychological Science, 2, 1-5.

Super, D.E. & Crites, J.O. (1965). Appraising vocational fitness by means of psychological

tests (Revised edition ed.). New York: Harper.

243

Taylor, T. R. (1994). A review of three approaches to cognitive assessment, and a proposed

integrated approach based on a unifying theoretical framework. South African Journal

of Psychology, 24(4), 184-157.

Theron, A. (2003). Perspectives on general and work behavior. In Z.C. Bergh & A.L.Theron

(Eds.), Psychology in the working context. (2nd ed., pp. 3-15). Cape Town: Oxford

University press Southern African.

Theron, C. (2009). The diversity-validity dilemma: In search of minimum adverse impact

and maximum utility. South African Journal of Industrial Psychology, 35(1), 1-13.

Thurstone, L. L. (1938). Primary mental abilities. London: University of Chicago press.

Tredoux, C. & Durrheim, K. (Eds.). (2002). Numbers, hypotheses and conclusions: A course

in statistics for social sciences. Landsdowne, Cape Town: UCT press.

Tsang, P. S. & Vidulich, M. A. (2003). Principles and practice of aviation psychology.

Mahwah, N.J.: Lawrence Erlbaum.

Van de Vijver, A. J.R. & Rothmann, S. (2004). Assessment in multicultural groups: The

South African case. South African Journal of Industrial Psychology, 30(4), 1-7.

Van den Berg, A. R. (1994). Intelligence tests. In K. Owen & J. J. Taljaard (Eds.), Handbook

for the use of psychological and scholastic tests of the IPER and NIPR . (83 – 136).

Pretoria: Human Sciences Research Council.

Van der Merwe, D. (2003). Predictive validity of a selection battery for technikon

students. (Unpublished Master’s dissertation). University of South Africa, Pretoria.

244

Van der Merwe, R. P. (1999). Psychological assessment in industry. South African Journal of

Industrial Psychology, 25(3), 8-11.

Van der Merwe, R. P. (2002). Psychometric testing and human resource

management. South African Journal of Industrial Psychology, 28(2), 77-86.

Van der Walt, H. S. (1998). Psychological tests: Refined instruments for reliable information

to decision-makers. In Focus Forum, 5(4), 16-17.

Van der Zanden, J. W. (1993). Human development (5th ed.). London: Oxford University

press.

Van Eeden, R. & De Beer, M. (2005). Assessment of cognitive functioning. In C. Foxcroft &

G. Roodt (Eds.), An introduction to psychological assessment in the South African

context (2nd Ed., pp. 119-139). Cape Town: Oxford university press, Southern Africa

(Pty) Ltd.

Van Eeden, R., De Beer, M. & Coetzee, C. H. (2001). Cognitive ability, learning potential,

and personality as predictors of academic achievement by engineering and other

science and technology students. South African Journal of Higher Education, 15(1),

171-179.

Van Rooyen, E. (2001). Die voorspelling van die akademisiese prestasie van studente in 'n

universitetsoorbruggingsprogram. South African Journal of Higher Education, 15(1),

180-188.

Vergnaud, G. (1997). Learning and teaching mathematics: An international perspective.

Hove, East Sussex: Psychology press.

245

Vermeulen, L.P. (2009). Flight instructor’s perception of pilot behavior related to gender.

South African Journal of Industrial Psychology. 35 (1), 1-8.

Vernon, P. E. (1969). Intelligence and cultural environment. London: Methen.

Viljoen, J. (1999). Education 121/EI. Johannesburg: Braamfontein College Publications.

Wagner, R. G. (1991). The psychology of reading: An introduction (2nd ed.). New York:

Oxford University press.

Walsh, W. & Betz, N. E. (1995). Tests and assessment. Englewood Cliffs: Prentice Hall.

Wechsler, D. (1997) Administration and scoring manual for the Wechsler Adult Intelligence

Scale-3rd Revision (WAIS-111). San Antonio: Psychological Corporation.

Wheeler, H. L. (1993). The fairness of an engineering selection battery in the mining

industry. (Unpublished Master’s dissertation). University of South Africa, Pretoria.

Wilcocks, A. (1973). Test administrator's Manual (Rev. ed.), Intermediate battery (B/77).

Johannesburg: Human Science Research Council.

Wilson, J. (2004). Gender - based issues in aviation, attitudes towards female pilots: A

cross - cultural analysis. (Unpublished Doctoral thesis). University of Pretoria, Pretoria.

Wray, D. & Medwell, J. (1993). Literacy and language in the primary years. London:

Routledge.

Zaaiman, H. (1998). Selecting students for mathematics and science: The challenge facing

higher education in South Africa. (Unpublished Doctoral thesis). Vrije Univeriteit,

Amsterdam.

246

Zevenbergen, R. (2001). Contexts in mathematics: Implications for literacy, literacy

learning, the middle years. A Journal of the Australian Literacy Educators

Association, 9(2), 21-28.

Zimmerman, C. H. (1957). Effective personnel selection procedures. London: Staple press.

Zindi, F. (1994). Differences in psychometric performance. The Psychologist, 7, 549-552.